Revisiting Design Criteria for Stormwater Treatment Systems Part One
Part I: Basins
It has been 25 years since the first community in the United States established the requirement for the post-development treatment of stormwater from new developments. Since then, local, regional, and state governments have published many manuals and handbooks identifying acceptable treatment systems (structural best management practices, or BMPs) and design criteria. Initially, few of these design criteria were supported by laboratory or field research. Absent such data, engineers were forced to use best professional judgment in choosing design criteria. With numerous studies completed over the past two decades, it is timely to reexamine some of these criteria.
Furthermore, our treatment strategy is rapidly evolving. The focus has been the general removal of pollutants using total suspended solids (TSS) as the surrogate for all pollutants. However, there has been a recent shift to consider specific pollutants. Some states now emphasize the removal of total phosphorus (e.g., Virginia, Washington, New York); total nitrogen (e.g., Maryland, North Carolina); and dissolved metals (e.g., Washington) in particular situations. With this more complex management strategy, a particular design criterion may differ depending upon the targeted pollutant.
This article is the first of a four-part series on design criteria. The first three articles consider three generic types of treatment systems: basins, fine-media filters, and vegetated swales and strips. The fourth article is devoted to the removal of dissolved pollutants. Differences in design criteria to target specific pollutants are also examined.
Table 1 provides design criteria for basins from representative manuals and handbooks. Design criteria related to size and configuration are discussed for extended detention ponds and vaults, wetponds and vaults, and constructed wetlands in their many variant forms. On the East Coast, the initial focus was on extended detention ponds, in some cases modified flood control facilities. These basins were expected to be empty after each storm. For many jurisdictions, the basic type has evolved to include some form of wet pool, either covering the entire bottom area or just a small pool, called a micropool, at the outlet. In contrast, on the West Coast, the starting point was wet basins, expected to retain stormwater between storms. However, some manuals now include an extended detention layer in the design. Convergence of concepts appears to be occurring, although the perspective differs. On the East Coast the perspective focuses on extended detention basins with wet pools; on the West Coast, on wet basins with an extended detention layer.
Basin Volume
In this first article, basin volume is considered with respect to the removal of sediment and attached pollutants such as metals and nutrients. The last article of this series considers the volume required to achieve a significant removal of dissolved pollutants.0.45
The volume of extended detention basins is commonly defined by specifying the depth of runoff to retain and treat. Initially, the specification was 0.5 inch, based on a philosophy of capturing the "first flush" of pollutants. It was reasoned that the majority of pollutants are present in the first 0.5 inch of runoff. Some jurisdictions began to state a management goal of capturing and treating "X" percent of the volume of stormwater over time, typically 85% - 90%. This led to an increase in runoff depth to 1 inch or more, as indicated in Table 1. With a volume capture goal of 85% - 90%, the relevance of first flush becomes problematic.
Not commonly recognized is that the sizing of an extended detention basin involves two questions: the depth of runoff to divert and the volume of the basin. These are two separate but related questions. It is common to assume that the volume of the basin is equal to the design depth of runoff that is to be diverted to achieve the management objective. However, for many jurisdictions, the volume of the basin must be larger than, not equal to, the runoff depth that is to be diverted. Why? It is important to recognize that some stormwater might at times remain in the basin as the next storm arrives. Hence, how much larger the basin must be depends on two factors: the design drawdown time and the interevent time, or the time between runoff events. The required basin volume increases as the interevent time decreases and as the specified drawdown time at brimful increases.
The above effects are understood from the work of Guo and Urbonas (1996) and related articles. Their methodology solves directly the relationship between the management goal and the size of the basin for a particular climatic region. The concept is illustrated in Figure 1. They proposed that the selected basin volume be equal to the maximized water-quality capture volume (WQCV), shown in Figure 1. The WQCV is defined as the point on the curve where an incremental increase in the storm volume captured and treated begins to decrease significantly with the incremental increase in basin volume. At the WQCV, the amount of stormwater captured over time is on the order of 80% - 95%, depending on the community, and is therefore close to the management goal of many communities. The normalized WQCV in Figure 1 is all events divided by a value of runoff depth equal to the 99.5 percentile runoff depth for each community. Normalization allows comparison between communities.
Representative results from Guo and Urbonas (1996) are presented in Table 2. Here the volume of the basin is specified as a multiple of the depth of the mean annual runoff event. The mean annual runoff event is the total annual runoff divided by the total number of runoff events (Driscoll et al. 1989). For example, assume a community has a mean annual storm runoff depth of 0.7 inches and a sizing ratio of 2. The volume of the extended detention basin is 1.4 inches times the tributary area and the runoff coefficient.
Table 2 shows the incremental effect on basin volume of increasing the drawdown time at brimful from 12 to 48 hours. It shows that increasing the drawdown time increases the required volume for the basin. This point is typically not recognized. Why must the volume be increased? Because the longer the drawdown time the more likely there will be stormwater in the basin when the next storm arrives. The effect differs significantly between communities. The incremental effect is minor for Boston, Tampa, and Denver. Conversely, it is significant for Seattle and Sacramento where the interevent times are relatively short during the wet season. For a given community, the aggregate volume of water captured over time is the same irrespective of the drawdown time specified at brimful. But the basin must be larger with greater design drawdown times. The ratios in Table 2 do not account for infiltration or evaporation.
For the specified storm depths presented in Table 1, the ratio of volume of the basin to the runoff volume of the mean annual event (Vb/Vr) ranges from 1.5 to 3. While the ratios lie within the vicinity of the WQCV values suggested in Table 2, it is unlikely values currently specified in many manuals that are the most appropriate for their regions. This is because, as noted previously, it is commonly assumed that the volume of the basin is equal to storm depth to be diverted.
Also presented in Table 2 are national values (ASCE 1998) based on the work of Guo. Given the substantial climatic differences between regions, apparent in Table 2, using average national values may result in either an undersized or oversized facility depending on the community.
Other methods for determining the WQCV have been proposed, such as the California Stormwater Quality Association (CASQA 2003); Goforth et al. (1983); King County (1998); Nix et al. (1983); Pitt, R. (2000); and Roesner et al. (1991). These methods have not been compared to ascertain whether they provide outcomes similar to Table 2. It is likely dissimilar results will be found. A professional dialogue is needed to ascertain the most appropriate method. For that reason, this author is not suggesting the approach by Guo and Urbonas (1996). Rather, it is presented to make the point that interevent time, the selected drawdown time, and the selected management objective interrelate to affect the design volume of a basin. As such, the selection of the drawdown time and the management objective are not separate decisions.
With respect to wet basins, Table 1 indicates that it is common practice to specify a volume that is the same as that specified for extended detention basins. Some manuals specify a larger volume. However, it is intuitive that a wet basin can be smaller than an extended detention basin if the desired removal efficiency or effluent concentration of TSS is the same. This is because additional settling occurs in wet basins between storm events. A synthesis of data of wet basin performance, Figure 2 suggests a Vb/Vr of 1 is sufficient (Strecker 2003). Figure 2 plots effluent concentration as a function of the Vb/Vr. The plot indicates little further reduction in the effluent concentration beyond a ratio of 1. Comparison of Figure 2 to Table 1 suggests that communities oversize wet pool basins if sediment and attached pollutants are the only objective. The data in Figure 2 reflect what is settling from the stormwater and the growth of plankton algae. Hence, TSS do not approach zero even with large basin volumes. It was previously observed with extended detention basins that maximization of the removal of particulate metals and nutrients likely requires a larger basin than if TSS removal is the sole objective. This may also be the case for wet basins. Hence, a Vb/Vr on the order of 1.5 may be desirable.
With respect to the West Coast perspective, the question arises as to the benefit of placing an extended detention layer atop a wet pool. Most manuals suggest splitting the design volume in half: half wet pool and half extended detention layer. A few suggest increasing the design volume to add an extended detention layer. The relative benefits of this approach are discussed later in this article.
Drawdown Time
Drawdown time is the time required to empty an extended detention basin or layer from brimful, the elevation of overflow. As noted previously, the design drawdown time at brimful affects the size of an extended detention basin; its specification is therefore important. Table 1 presents a significant range of drawdown times, from 24 to 72 hours at brimful.
The first recommendation for drawdown time was 40 hours at brimful, averaging 24 hours for all events (Grizzard et al. 1987). The recommendation was derived from a field study of one converted flood control detention pond and laboratory column tests. The laboratory studies showed little additional removal of TSS beyond a settling time of six to 24 hours depending on the initial TSS concentration. A more recent study of a field basin (Keblin et al. 1998) found negligible improvement in TSS reduction at drawdown times in excess of 24 hours. As settling velocity distributions of sediment in stormwater can vary regionally because of varying soil types, the appropriate drawdown time may also vary regionally. For example, in New England, where a significant fraction of the sediment may be deicing sand, a lower drawdown time is likely needed than in the Pacific Northwest to achieve the same performance goal. Other pollutants need to be considered. Generally, metals and phosphorus associate primarily with the fine sediments, necessitating relatively longer drawdown times to achieve high removal of the particulate forms. The work of Keblin et al. (1998) suggests that achieving the highest practical removal efficiency of particulate metals may be twice that of only the sediment.
It appears that the average drawdown time should be at least 24 hours. However, 48 hours is likely prudent in most regions. This results in drawdown times at brimful on the order of 40 to 72 hours. There is little reason not to specify a greater drawdown time in communities, such as Boston, where the effect on basin volume is minor. If the effect of the decision is significant, as with Seattle and Sacramento, local studies of the relationship between performance and drawdown time should be conducted. This suggests that the analysis of the relationship between drawdown time at brimful and basin volume needs to be extended beyond 48 hours.
The above discussion is relevant to extended detention basins that dry completely between storms. However, most manuals now encourage inclusion of some form of permanent wet pool. This raises the question as to whether very long drawdown times are necessary. One manufactured product, the StormVault, uses the combined wet pool/extended detention layer configuration. It appears to provide satisfactory treatment with a drawdown time at brimful of only six hours. The contribution of the wet pool to performance is likely significant.
Drawdown Rate Although drawdown time has been in use for two decades, it is not necessarily the correct design criterion. The more appropriate criterion may be the drawdown rate; that is, the rate at which the water level drops in the basin. Consider two basins, identified as A and B, both with the same design volume and drawdown time at brimful. They differ in that Basin A has an average depth at brimful one-third that of Basin B. Hence, Basin A has three times the surface area of Basin B. It follows that the average drawdown rate of Basin A is one-third that of Basin B. As a consequence, Basin A removes particles that are much smaller, with lower settling velocities, than Basin B. Conceptually, the perspective is to recognize that all particles with settling velocities greater than the fall rate of the water will reach the bottom of the basin before the water exits the basin. The slower the fall rate, the smaller the particle that reaches the bottom. The fall rate in an extended detention basin is akin to the hydraulic loading rate (HLR). The HLR is expressed as cubic feet per second per square foot of basin surface area. It will be noted that the units of cfs/ft2 are also ft/sec: the units of the settling velocity of a particle. The relevance of the HLR has been recognized in water and wastewater treatment for a century (Hazen 1904) and is the basis for sizing sedimentation basins (AWWA 1990; Metcalf and Eddy 1991). Its appropriateness has also been recognized for stormwater treatment (Small and DiToro 1979; USEPA, 1986; Dorman et al. 1996).
Is there a benefit of specifying drawdown rate rather than time? Table 2 suggests it depends upon the community. The advantage of specifying the drawdown rate is that the needed drawdown time and therefore basin volume decreases with a shallower basin, compensating for the larger area required of a shallower basin. For example, assume the local jurisdiction selects a drawdown rate of 0.15 ft/hr. For a basin with a depth of 7.2 feet, the drawdown time is 48 hours at brimful. If a depth of 3.6 feet is selected, the drawdown time need only be 24 hours. According to Table 2, the required basin volume is only about 5% less for Boston and Cincinnati, 10% for Denver and Tampa, but approximately 40% for Seattle and Sacramento. What is the appropriate drawdown rate? Figure 3 summarizes information for several studies of extended detention basins. The data are from facilities in California, Maryland, North Carolina, Virginia, and Texas. Shown is the effluent concentration versus drawdown rate. Basing the choice of drawdown rate on effluent concentration may be more valid than efficiency, because efficiency tends to be lower at lower influent concentrations. However, effluent concentration is not independent of influent concentration, complicating the analysis.
As the data are scattered, the appropriate drawdown rate is not readily obvious. Some of the scatter is due to highly varying median influent concentrations. It is also noted that the performances observed in each study reflect drawdown rates that are much lower than that at brimful because most storms in each study did not fill the respective basin. Given the scatter in Figure 3, one might conclude we should continue to use drawdown time. However, to be fair, if the abscissa in Figure 3 were plotted against drawdown time, a similar scatter would be found. It would be just as difficult to select a drawdown time from such a scatter. Figure 3 suggests that below 0.30 ft/hr, the incremental benefit is problematic. Based on the work of Keblin et al. (1998) previously cited, this value should be halved to 0.15 ft/hr to achieve the maximum practical removal of attached pollutants. This value represents the average drawdown rate from brimful
The size of particle that settles faster than a drawdown rate of 0.15 ft/hr is about 5 to 15 microns, depending on water temperature and the specific gravity and shape of the particle. However, particles smaller than this size range will be found on the bottom of the basin for two reasons. First, during all events the fall rate is for much of the time less than the average rate at brimful. Secondly, the stormwater as it enters the basin is thoroughly mixed vertically. As a consequence, many particles less than the 5- to 15-micron range reach the bottom before the basin empties because they are close to the bottom as they enter the basin.
Is detention time irrelevant? Drawdown rate is relevant to particles that are discrete. That is, as they fall, coagulation does not occur when two particles make contact. Silts and sands are discrete suspensions, whereas clays coagulate given sufficient time. By convention, clay particles are smaller than 5 microns. Hence, at a drawdown rate of 0.15 ft/hr, much of the clay suspension might not reach the bottom before the basin empties unless it coagulates into larger particles. It is possible at influent sediment concentrations less than about 50 mg/L the clay fraction is significant, in which case time for coagulation may be important. Grizzard et al. (1986) found at these low concentrations it took on the order of 48 hours to reach a percent removal similar to that achieved in only two hours when the initial TSS concentration was on the order of 100 mg/L. This suggests extended time is needed for coagulation at very low initial concentrations. However, it is questionable as to whether high levels of performance are relevant at such low initial concentrations. Some manuals specify a drawdown time of 24 hours for the average condition, but this average condition is not typically defined. Regardless, it is likely that even with this specification, clays have insufficient time to satisfactorily coagulate. Hence, there may be an inherent limitation for extended detention basins if clay is a significant fraction of the incoming sediment. As such the benefit of a wet pool becomes apparent. The long residence time between runoff events provides sufficient time for the clay to coagulate in the wet pool and to reach the bottom of the basin. These issues again point to the importance of understanding local conditions and influent concentrations when selecting an appropriate drawdown rate.
Length-to-Width Ratio The purpose of specifying a length-to-width (L/W) ratio is to improve hydraulic efficiency by reducing short-circuiting and dead zones. With respect to wet basins, hydraulic efficiency is defined as how well fresh stormwater exchanges with older water in the basin. Table 1 indicates a large range in design values. Specifying an L/W ratio for extended detention basins is likely unnecessary as these basins operate essentially as fill-and-draw systems. The constricted outlet tends to force incoming stormwater into all areas of the basin. A comparison of six extended detention basins in which the L/W ratio varied from 3 to 10 found no difference in performance (Caltrans 2004). Whether performance is affected by ratios in the range of 1 to 3 is unknown.
While a high L/W ratio may maximize hydraulic efficiency, it may also create a geometry that is difficult to fit into many developments. Furthermore, excessively narrow basins may induce the resuspension of sediments because of greater throughput water velocities. The experience from wastewater treatment is to have substantial L/W ratios, on the order 10 to 20 or greater depending on the unit operation (e.g., sedimentation, chlorination). Such high L/W ratios are not necessary in stormwater systems because the rate of inflow is generally lower relative to the basin volume than in wastewater systems.
One study of the relationship between the L/W ratio and hydraulic efficiency suggests a ratio above 2 provides little additional benefit. Figure 4 is an interpretation of Walker (1998). It shows the effect of increasing the L/W ratio on hydraulic efficiency. Figure 4 suggests modest benefit of incremental increases of the ratio above 2. It also suggests the appropriate L/W ratio likely differs with the size of the basin as defined by the Vb/Vr value. For basins with large Vb/Vr values a L/W ratio of 2 is likely satisfactory. If the Vb/Vr of wet basins is on the order of 1 as recommended in this article, the L/W ratio should be greater, perhaps 3 to 4. The storm intensity common to the region may also be relevant to this decision. In the Southeast, with high-intensity storms, the incremental benefit of increasing the L/W ratio is likely greater than the Pacific Northwest with its mild storms. This difference is suggested by contrasting the results of Persson et al. (1999) and Persson (2000) to Walker (1996). In the former studies significant benefit was found up to a ratio of 5. However, these studies were conducted at higher flow rates.
Poor hydraulic efficiency has been found not to be an issue with poorly configured large wet basins with Vb/Vr values on the order of 10 (Pitt 2000). This is not likely the case with smaller Vb/Vr ratios. Baffles were added to a pond with a Vb/Vr of about 1.5, increasing the L/W from 1.5 to 4.5 (Mathews et al. 1997). A significant improvement in hydraulic efficiency was observed.
An L/W ratio of 1 is likely valid for small wet basins if an extended detention layer is included on top of the wet pool. Restricting the outlet causes stormwater to "back into" areas that would otherwise be dead zones such as corners or areas of dense vegetation. The designer may be allowed the flexibility of using either a wet pool only with a large L/W ratio, or a combination extended detention/wet pool with a smaller L/W ratio to meet site constraints. For pure extended detention basins a L/W ratio of 1 is likely sufficient.
As with the L/W ratio, the benefits of an extended detention layer may differ with the climatic region. The concept may be most beneficial in the Southeast with its short, high-intensity storms and high rates of inflow. In contrast, the concept may be of less benefit on the West Coast where average storm intensities are much less.
The above discussion does not take into consideration the potential adverse effects of thermal or density (from deicing salts) stratification on hydraulic and therefore performance efficiency. The effect of stratification is poorly understood. Thermal stratification reduced the detention time of a small wet basin by half (Timmins et al. 1999). The significance of stratification to performance is unknown, as is the effect of the energy of incoming storms to temporarily destratify the basin. One study concluded that summer storms abetted stratification due to the incoming stormwater being higher in temperature than the pond water (McBean and Burn 1983).
Stratification should not reduce performance during storms. As established above, performance is a function of hydraulic loading rate, which remains the same irrespective of stratification. However, stratification reduces the exchange of fresh stormwater with water present in the basin. As a consequence, stratification in effect reduces performance between storms with respect to fine sediments and dissolved pollutants. Increasing the L/W ratio might decrease the potential for stratification by increasing flow velocities within the basin. Subsurface discharge to the lower half of the wet pool and/or entrance baffles might have a greater mitigating effect than a large L/W. A final note of interest is that in the study cited above, thermal stratification occurred in the top 10 inches of the wet pool. This suggests that the common recommendation (Table 1) to limit the maximum depth to avoid stratification may not be effective. More studies on this subject are needed.
Multiple Ponds or Cells Several manuals suggest incorporating multiple cells, either as separate basins in series or by separation of one basin into two or more cells with a berm extending above the water surface. Multiple cells can significantly increase space requirements. The purported benefit is more effective treatment. However, there are no data substantiating this claim. Some conceptual designs like that shown in Figure 5 may result in poorer performance than one basin of equal volume. While aesthetically pleasing, multi-ponds with irregular shapes increase short-circuiting and dead zones. This decreases hydraulic efficiency and, as a consequence, performance efficiency. An extended detention layer may minimize this effect.
Vegetation Coverage
For wetponds, a shallow safety bench of 10 to 15 feet in width is commonly recommended. The benches become covered by emergent vegetation. For wetlands, manuals usually specify a depth regime that results in vegetation coverage on the order 60% - 85% of the surface area. Concepts of vegetation configurations are frequently included in manuals (Figure 5).
Depending on its degree and location, emergent vegetation may enhance or degrade hydraulic efficiency and therefore performance efficiency. If the basin is configured so as to produce an open area down the middle of the basin, hydraulic efficiency and therefore performance is reduced. Water, finding the path of least resistance, moves through the center of the basin, with little exchange with "old" water in the densely vegetated areas along the sides of the basin (Persson et al. 1999). In shallow marsh wetlands, low-flow channels may have a similar effect. As the wetland ages, a pattern of uneven plant densities may create channels of less resistance, decreasing hydraulic efficiency.
A more appropriate configuration might be open fore- and after bays with a shallow intermediate marsh area and no low-flow channel. To use less space, the intermediate area could be of wetpond depth. The interface between the central area and the fore- or after bays can be a berm, but with the top below the water and covered by emergent vegetation. This berm configuration could more evenly spread incoming water across the lateral cross-section of the pond. Alternatively, as previously noted, an extended detention layer might compensate for the adverse effects of fringe vegetation on hydraulic efficiency.
Sediment Storage
The common criterion is to either add 1 foot of depth or increase the basin volume by 20% above that calculated for performance. The added cost is not trivial. However, once construction of the development is complete, the accumulation rate of sediment is modest, on the order of 0.25 to 0.5 in/yr. Furthermore, most of the incoming sediment settles in the forebay. It is suggested that this requirement be eliminated, particularly if a forebay is inherent to the design. Furthermore, a more cost-effective alternative to the earthen forebay might be a large standard manhole, a small manufactured wet vault (e.g., Stormceptor), or a vortex separator. Even if the initial cost is greater, the life-cycle cost might be lower given the greater ease of maintenance of these devices.
Final Observations
The discussion suggests that wet basins and extended detention basins are not separate types but opposite ends of the same type. A continuum exists between the two extremes, depending on the climate of the particular region and the design volume. The discussion further suggests that extended detention basins should perform as well as wet basins, although they must be larger to achieve similar performance with respect to the removal of sediment and particulate pollutants. As previously noted, particles as small as 5 to 10 microns should be readily removed at current drawdown rates. However, performance studies suggest otherwise, as implied in Figure 3. TSS in the effluent of wetponds are typically on the order of 10 to 20 mg/L, whereas they are on the order of 20 to 40 mg/L for extended detention basins. This discrepancy is due in part to wetponds having volumes equal to or greater than those of extended detention basins (Table 1). It may also be due in part to the likely inability to induce the coagulation of clay even with a specified drawdown time for small storms.
Design aspects other than volume and drawdown time might contribute to a less-than-expected performance for extended detention basins. The most likely reasons are adverse inlet and outlet conditions. Resuspension of previously deposited sediment and erosion may occur in the inlet area during the infrequent high-intensity storms if energy dissipation is insufficient. Resuspended clays might have insufficient time to re-coagulate and as a result exit during these events. The more likely candidate for subpar performance is the outlet structure. In a water or wastewater basin, exiting water is gathered along an effluent weir that extends the width of the basin. This minimizes approach velocities to the weir. By contrast, the outlet structures of extended detention basins are very small relative to their end-cross sections. As a consequence, stormwater "rushes" toward one relatively small point in the basin. Approach velocities near the outlet are excessive relative to the settling velocity of silts and clays, and even fine sand. The effective HLR in the vicinity of the outlet is substantially greater than the nominal HLR of the basin. Sediments that should be removed based on the nominal HLR, or drawdown rate, are lost through the outlet. This condition is exacerbated if the bottom around the outlet is bare, allowing for erosion at the high approach velocities. While wet basins also have narrow outlets, the effect is less severe. With wet basins the water exiting each storm is "old" water that is relatively clear of fine sediments, given the retention time between storms.
One or a combination of design elements may mitigate the effect of small outlet structures. Several outlet risers tied to an exit manifold may help. However, the issue of clogging very small orifices arises. A shallow wet pool or micropool at the outlet may counter the effect of excessive approach velocities by preventing scouring of accumulated sediments at the bottom of the basin in the vicinity of the riser structure (Fennessey and Jarrett 1997). Some have found that withdrawal at the surface improves performance (Ward et al. 1979; Millen et al. 1997) whereas others have not (Keblin et al. 1998). Spacing, size, and the location of perforations on the riser may affect performance (Ward et al. 1979). However, another study found no significant difference in performance between a single bottom orifice and a perforated vertical riser (Fennessey and Jarrett 1997). It is possible that while the riser configuration reduces approach velocities, the reduction may not be sufficient to significantly affect performance. Certainly, however, a riser structure with a series of vertically placed holes should reduce erosion and/or resuspension in the immediate vicinity of the outlet. A cone of gravel around the bottom of the outlet structure has been found to significantly improve performance (Engle and Jarrett 1995). More studies on this issue are needed. Basins sized as suggested here should be able to meet the general criterion of 80% removal of TSS, a specification of most manuals. They should also be able to meet the specification of 40% removal of the total phosphorus (TP) inasmuch as 75% - 90% of the phosphorus is typically bound to the sediments. Meeting the goal of 40% removal of nitrogen is less certain given that it is more soluble than phosphorus. These questions are considered in the fourth article of this series.
It is common to specify a minimum drainage area, under the thesis that a base flow of water into the basin is needed to keep a permanent pool. This limits the use of wet basins for larger areas, on the order of at least 10 to 25 acres depending on the region. The concern is that the wet basin will not fully function without water. This is not the case. There is no need for the basin to "work" between storms. Pollutants migrate to the bottom as water evaporates and into the soil with infiltration. Allowing the pond to dry may reduce mosquito problems. Infiltration is good for the hydrologic cycle. However, drying may solubilize some pollutants, which might be lost in subsequent storms. If the removal target is solely TSS and attached pollutants, this is no concern. Even if dissolved pollutants are of concern, the pond or wetland fills with the next storm, at which time the solubilized pollutants likely renew attachment to inorganic and organic sediments. Constructed wetlands can withstand extended periods of dryness without harm to the vegetation. Specifications as to the maximum length of dryness to avoid this problem may be prudent and will vary with the climatic region and vegetation species. It is also important to recognize that a constant base flow can reduce the performance with some pollutants. One study found a negative net removal of phosphorus (Oberts 1997). Removal during storms was offset by loss in base flows during dry periods.
The concepts discussed in this article are also relevant to manufactured basins such as Stormceptor, BaySaver, EcoStorm, and Stormvault. Regardless of how they may be configured, they are essentially wet vaults.
Summary
The data suggest wet basins can be smaller than currently specified. They also can be smaller than extended detention basins if the sole objective is the reduction of sediment and attached pollutants such as metals, pesticides, and nutrients. In most situations this "basic" level of treatment is likely sufficient. A dead storage volume equal to the runoff of the mean annual event‹that is, a Vb/Vr of 1 to 1.5‹is likely adequate for wet basins. A larger volume may be necessary where significant removal of dissolved pollutants is also desired. This aspect is considered in the fourth article of this series.
To achieve reasonably equivalent performance, extended detention basins must be larger than wet basins with a Vb/Vr on the order of 2 to 2.5, depending on the region. The specification of volume must also take into consideration the specified drawdown time at brimful and the interevent time for the locality. A longer specified drawdown time and shorter interevent time requires a larger basin volume to achieve the same management objective with respect to the aggregate volume of water treated over time. The effect of infiltration and evaporation could be included in this analysis.
Drawdown rate rather than drawdown time is the correct design criterion for extended detention basins. In some regions, in particular the West Coast, recognition of drawdown rate as the correct criterion results in more cost-effective sizing. Current data suggest the specification for the average drawdown rate from brimful be on the order of 0.15 ft/hr. Although there is considerable uncertainty with this specification, the same uncertainty exists with the current specification for drawdown time. Regardless, even if changing from drawdown time to rate has little effect on basin volumes for much of the nation, the use of drawdown rate will ensure that field studies gather the appropriate information that has been lacking in most previous studies.
Drawdown time may be relevant to the extent that clay is a significant fraction of the incoming sediment. Local studies are needed to ascertain the relevance of clay as are more studies on the relationship between time, coagulation, and settling. The data suggest, however, that the issue of clay might not be the most significant factor in what appears to be less-than-expected performance of dry basins.
There may be an inherent limitation of extended detention basins to perform as well as wet basins irrespective of the drawdown rate or time. Design aspects other than volume and drawdown time must be considered, in particular the outlet structure. Modifications to current outlet design specifications are needed to reduce approaching velocities and the potential for resuspension and/or erosion in the vicinity of the outlet. The limitations of these basins can be mitigated through the inclusion of a wet pool. However, at some point the question arises as to what is being designed: an extended detention basin with a wet pool or a wet basin with an extended detention layer. It follows that at some point the relevance of an extended detention layer becomes problematic.
The length-to-width ratio is an important design parameter for wet basins, but it is probably not particularly relevant to extended detention basins. A ratio on the order of 2 to 4 is likely sufficient for wet basins. The appropriate ratio depends on the selected Vb/Vr and general rainfall intensity. A ratio of 1 is likely sufficient for extended detention basins. Oddly configured basins or basins with varying bathymetry adversely affect hydraulic efficiency and therefore performance efficiency. The adverse effect of such configurations can be reduced by increasing the size of the system and/or by including an extended detention layer. However, as these relationships are little understood, it is advisable to carry out the appropriate studies before imposing such requirements.
Advertisement
Care must be taken with emergent vegetation. Depending on location and density, emergent vegetation can be counterproductive. Vegetation along the sides of a wet basin, with open water down the center, reduces hydraulic efficiency. This can be countered by lateral berms with emergent vegetation and/or an extended detention layer. The significance of this observation is greater in regions with high-intensity storms such as the Southeast.
Additional laboratory and field studies are needed to better define the significance of the observations and recommendations made in this article. Being aware of the relevant design criteria and how their selection may affect performance leads to studies in which the most relevant characteristics of performance are evaluated such as hydraulic efficiency, water level fluctuations, and the size distributions of incoming and outgoing sediments.
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.
November-December 2004
Revisiting Design Criteria for Stormwater Treatment Systems Part One
Part I: Basins
It has been 25 years since the first community in the United States established the requirement for the post-development treatment of stormwater from new developments. Since then, local, regional, and state governments have published many manuals and handbooks identifying acceptable treatment systems (structural best management practices, or BMPs) and design criteria. Initially, few of these design criteria were supported by laboratory or field research. Absent such data, engineers were forced to use best professional judgment in choosing design criteria. With numerous studies completed over the past two decades, it is timely to reexamine some of these criteria.
Furthermore, our treatment strategy is rapidly evolving. The focus has been the general removal of pollutants using total suspended solids (TSS) as the surrogate for all pollutants. However, there has been a recent shift to consider specific pollutants. Some states now emphasize the removal of total phosphorus (e.g., Virginia, Washington, New York); total nitrogen (e.g., Maryland, North Carolina); and dissolved metals (e.g., Washington) in particular situations. With this more complex management strategy, a particular design criterion may differ depending upon the targeted pollutant.
This article is the first of a four-part series on design criteria. The first three articles consider three generic types of treatment systems: basins, fine-media filters, and vegetated swales and strips. The fourth article is devoted to the removal of dissolved pollutants. Differences in design criteria to target specific pollutants are also examined.
Table 1 provides design criteria for basins from representative manuals and handbooks. Design criteria related to size and configuration are discussed for extended detention ponds and vaults, wetponds and vaults, and constructed wetlands in their many variant forms. On the East Coast, the initial focus was on extended detention ponds, in some cases modified flood control facilities. These basins were expected to be empty after each storm. For many jurisdictions, the basic type has evolved to include some form of wet pool, either covering the entire bottom area or just a small pool, called a micropool, at the outlet. In contrast, on the West Coast, the starting point was wet basins, expected to retain stormwater between storms. However, some manuals now include an extended detention layer in the design. Convergence of concepts appears to be occurring, although the perspective differs. On the East Coast the perspective focuses on extended detention basins with wet pools; on the West Coast, on wet basins with an extended detention layer.
Basin Volume
In this first article, basin volume is considered with respect to the removal of sediment and attached pollutants such as metals and nutrients. The last article of this series considers the volume required to achieve a significant removal of dissolved pollutants.0.45
The volume of extended detention basins is commonly defined by specifying the depth of runoff to retain and treat. Initially, the specification was 0.5 inch, based on a philosophy of capturing the "first flush" of pollutants. It was reasoned that the majority of pollutants are present in the first 0.5 inch of runoff. Some jurisdictions began to state a management goal of capturing and treating "X" percent of the volume of stormwater over time, typically 85% - 90%. This led to an increase in runoff depth to 1 inch or more, as indicated in Table 1. With a volume capture goal of 85% - 90%, the relevance of first flush becomes problematic.
Not commonly recognized is that the sizing of an extended detention basin involves two questions: the depth of runoff to divert and the volume of the basin. These are two separate but related questions. It is common to assume that the volume of the basin is equal to the design depth of runoff that is to be diverted to achieve the management objective. However, for many jurisdictions, the volume of the basin must be larger than, not equal to, the runoff depth that is to be diverted. Why? It is important to recognize that some stormwater might at times remain in the basin as the next storm arrives. Hence, how much larger the basin must be depends on two factors: the design drawdown time and the interevent time, or the time between runoff events. The required basin volume increases as the interevent time decreases and as the specified drawdown time at brimful increases.
The above effects are understood from the work of Guo and Urbonas (1996) and related articles. Their methodology solves directly the relationship between the management goal and the size of the basin for a particular climatic region. The concept is illustrated in Figure 1. They proposed that the selected basin volume be equal to the maximized water-quality capture volume (WQCV), shown in Figure 1. The WQCV is defined as the point on the curve where an incremental increase in the storm volume captured and treated begins to decrease significantly with the incremental increase in basin volume. At the WQCV, the amount of stormwater captured over time is on the order of 80% - 95%, depending on the community, and is therefore close to the management goal of many communities. The normalized WQCV in Figure 1 is all events divided by a value of runoff depth equal to the 99.5 percentile runoff depth for each community. Normalization allows comparison between communities.
Representative results from Guo and Urbonas (1996) are presented in Table 2. Here the volume of the basin is specified as a multiple of the depth of the mean annual runoff event. The mean annual runoff event is the total annual runoff divided by the total number of runoff events (Driscoll et al. 1989). For example, assume a community has a mean annual storm runoff depth of 0.7 inches and a sizing ratio of 2. The volume of the extended detention basin is 1.4 inches times the tributary area and the runoff coefficient.
Table 2 shows the incremental effect on basin volume of increasing the drawdown time at brimful from 12 to 48 hours. It shows that increasing the drawdown time increases the required volume for the basin. This point is typically not recognized. Why must the volume be increased? Because the longer the drawdown time the more likely there will be stormwater in the basin when the next storm arrives. The effect differs significantly between communities. The incremental effect is minor for Boston, Tampa, and Denver. Conversely, it is significant for Seattle and Sacramento where the interevent times are relatively short during the wet season. For a given community, the aggregate volume of water captured over time is the same irrespective of the drawdown time specified at brimful. But the basin must be larger with greater design drawdown times. The ratios in Table 2 do not account for infiltration or evaporation.
For the specified storm depths presented in Table 1, the ratio of volume of the basin to the runoff volume of the mean annual event (Vb/Vr) ranges from 1.5 to 3. While the ratios lie within the vicinity of the WQCV values suggested in Table 2, it is unlikely values currently specified in many manuals that are the most appropriate for their regions. This is because, as noted previously, it is commonly assumed that the volume of the basin is equal to storm depth to be diverted.
Also presented in Table 2 are national values (ASCE 1998) based on the work of Guo. Given the substantial climatic differences between regions, apparent in Table 2, using average national values may result in either an undersized or oversized facility depending on the community.
Other methods for determining the WQCV have been proposed, such as the California Stormwater Quality Association (CASQA 2003); Goforth et al. (1983); King County (1998); Nix et al. (1983); Pitt, R. (2000); and Roesner et al. (1991). These methods have not been compared to ascertain whether they provide outcomes similar to Table 2. It is likely dissimilar results will be found. A professional dialogue is needed to ascertain the most appropriate method. For that reason, this author is not suggesting the approach by Guo and Urbonas (1996). Rather, it is presented to make the point that interevent time, the selected drawdown time, and the selected management objective interrelate to affect the design volume of a basin. As such, the selection of the drawdown time and the management objective are not separate decisions.
With respect to wet basins, Table 1 indicates that it is common practice to specify a volume that is the same as that specified for extended detention basins. Some manuals specify a larger volume. However, it is intuitive that a wet basin can be smaller than an extended detention basin if the desired removal efficiency or effluent concentration of TSS is the same. This is because additional settling occurs in wet basins between storm events. A synthesis of data of wet basin performance, Figure 2 suggests a Vb/Vr of 1 is sufficient (Strecker 2003). Figure 2 plots effluent concentration as a function of the Vb/Vr. The plot indicates little further reduction in the effluent concentration beyond a ratio of 1. Comparison of Figure 2 to Table 1 suggests that communities oversize wet pool basins if sediment and attached pollutants are the only objective. The data in Figure 2 reflect what is settling from the stormwater and the growth of plankton algae. Hence, TSS do not approach zero even with large basin volumes. It was previously observed with extended detention basins that maximization of the removal of particulate metals and nutrients likely requires a larger basin than if TSS removal is the sole objective. This may also be the case for wet basins. Hence, a Vb/Vr on the order of 1.5 may be desirable.
With respect to the West Coast perspective, the question arises as to the benefit of placing an extended detention layer atop a wet pool. Most manuals suggest splitting the design volume in half: half wet pool and half extended detention layer. A few suggest increasing the design volume to add an extended detention layer. The relative benefits of this approach are discussed later in this article.
Drawdown Time
Drawdown time is the time required to empty an extended detention basin or layer from brimful, the elevation of overflow. As noted previously, the design drawdown time at brimful affects the size of an extended detention basin; its specification is therefore important. Table 1 presents a significant range of drawdown times, from 24 to 72 hours at brimful.
The first recommendation for drawdown time was 40 hours at brimful, averaging 24 hours for all events (Grizzard et al. 1987). The recommendation was derived from a field study of one converted flood control detention pond and laboratory column tests. The laboratory studies showed little additional removal of TSS beyond a settling time of six to 24 hours depending on the initial TSS concentration. A more recent study of a field basin (Keblin et al. 1998) found negligible improvement in TSS reduction at drawdown times in excess of 24 hours. As settling velocity distributions of sediment in stormwater can vary regionally because of varying soil types, the appropriate drawdown time may also vary regionally. For example, in New England, where a significant fraction of the sediment may be deicing sand, a lower drawdown time is likely needed than in the Pacific Northwest to achieve the same performance goal. Other pollutants need to be considered. Generally, metals and phosphorus associate primarily with the fine sediments, necessitating relatively longer drawdown times to achieve high removal of the particulate forms. The work of Keblin et al. (1998) suggests that achieving the highest practical removal efficiency of particulate metals may be twice that of only the sediment.
It appears that the average drawdown time should be at least 24 hours. However, 48 hours is likely prudent in most regions. This results in drawdown times at brimful on the order of 40 to 72 hours. There is little reason not to specify a greater drawdown time in communities, such as Boston, where the effect on basin volume is minor. If the effect of the decision is significant, as with Seattle and Sacramento, local studies of the relationship between performance and drawdown time should be conducted. This suggests that the analysis of the relationship between drawdown time at brimful and basin volume needs to be extended beyond 48 hours.
The above discussion is relevant to extended detention basins that dry completely between storms. However, most manuals now encourage inclusion of some form of permanent wet pool. This raises the question as to whether very long drawdown times are necessary. One manufactured product, the StormVault, uses the combined wet pool/extended detention layer configuration. It appears to provide satisfactory treatment with a drawdown time at brimful of only six hours. The contribution of the wet pool to performance is likely significant.
Drawdown Rate Although drawdown time has been in use for two decades, it is not necessarily the correct design criterion. The more appropriate criterion may be the drawdown rate; that is, the rate at which the water level drops in the basin. Consider two basins, identified as A and B, both with the same design volume and drawdown time at brimful. They differ in that Basin A has an average depth at brimful one-third that of Basin B. Hence, Basin A has three times the surface area of Basin B. It follows that the average drawdown rate of Basin A is one-third that of Basin B. As a consequence, Basin A removes particles that are much smaller, with lower settling velocities, than Basin B. Conceptually, the perspective is to recognize that all particles with settling velocities greater than the fall rate of the water will reach the bottom of the basin before the water exits the basin. The slower the fall rate, the smaller the particle that reaches the bottom. The fall rate in an extended detention basin is akin to the hydraulic loading rate (HLR). The HLR is expressed as cubic feet per second per square foot of basin surface area. It will be noted that the units of cfs/ft2 are also ft/sec: the units of the settling velocity of a particle. The relevance of the HLR has been recognized in water and wastewater treatment for a century (Hazen 1904) and is the basis for sizing sedimentation basins (AWWA 1990; Metcalf and Eddy 1991). Its appropriateness has also been recognized for stormwater treatment (Small and DiToro 1979; USEPA, 1986; Dorman et al. 1996).
Is there a benefit of specifying drawdown rate rather than time? Table 2 suggests it depends upon the community. The advantage of specifying the drawdown rate is that the needed drawdown time and therefore basin volume decreases with a shallower basin, compensating for the larger area required of a shallower basin. For example, assume the local jurisdiction selects a drawdown rate of 0.15 ft/hr. For a basin with a depth of 7.2 feet, the drawdown time is 48 hours at brimful. If a depth of 3.6 feet is selected, the drawdown time need only be 24 hours. According to Table 2, the required basin volume is only about 5% less for Boston and Cincinnati, 10% for Denver and Tampa, but approximately 40% for Seattle and Sacramento. What is the appropriate drawdown rate? Figure 3 summarizes information for several studies of extended detention basins. The data are from facilities in California, Maryland, North Carolina, Virginia, and Texas. Shown is the effluent concentration versus drawdown rate. Basing the choice of drawdown rate on effluent concentration may be more valid than efficiency, because efficiency tends to be lower at lower influent concentrations. However, effluent concentration is not independent of influent concentration, complicating the analysis.
As the data are scattered, the appropriate drawdown rate is not readily obvious. Some of the scatter is due to highly varying median influent concentrations. It is also noted that the performances observed in each study reflect drawdown rates that are much lower than that at brimful because most storms in each study did not fill the respective basin. Given the scatter in Figure 3, one might conclude we should continue to use drawdown time. However, to be fair, if the abscissa in Figure 3 were plotted against drawdown time, a similar scatter would be found. It would be just as difficult to select a drawdown time from such a scatter. Figure 3 suggests that below 0.30 ft/hr, the incremental benefit is problematic. Based on the work of Keblin et al. (1998) previously cited, this value should be halved to 0.15 ft/hr to achieve the maximum practical removal of attached pollutants. This value represents the average drawdown rate from brimful
The size of particle that settles faster than a drawdown rate of 0.15 ft/hr is about 5 to 15 microns, depending on water temperature and the specific gravity and shape of the particle. However, particles smaller than this size range will be found on the bottom of the basin for two reasons. First, during all events the fall rate is for much of the time less than the average rate at brimful. Secondly, the stormwater as it enters the basin is thoroughly mixed vertically. As a consequence, many particles less than the 5- to 15-micron range reach the bottom before the basin empties because they are close to the bottom as they enter the basin.
Is detention time irrelevant? Drawdown rate is relevant to particles that are discrete. That is, as they fall, coagulation does not occur when two particles make contact. Silts and sands are discrete suspensions, whereas clays coagulate given sufficient time. By convention, clay particles are smaller than 5 microns. Hence, at a drawdown rate of 0.15 ft/hr, much of the clay suspension might not reach the bottom before the basin empties unless it coagulates into larger particles. It is possible at influent sediment concentrations less than about 50 mg/L the clay fraction is significant, in which case time for coagulation may be important. Grizzard et al. (1986) found at these low concentrations it took on the order of 48 hours to reach a percent removal similar to that achieved in only two hours when the initial TSS concentration was on the order of 100 mg/L. This suggests extended time is needed for coagulation at very low initial concentrations. However, it is questionable as to whether high levels of performance are relevant at such low initial concentrations. Some manuals specify a drawdown time of 24 hours for the average condition, but this average condition is not typically defined. Regardless, it is likely that even with this specification, clays have insufficient time to satisfactorily coagulate. Hence, there may be an inherent limitation for extended detention basins if clay is a significant fraction of the incoming sediment. As such the benefit of a wet pool becomes apparent. The long residence time between runoff events provides sufficient time for the clay to coagulate in the wet pool and to reach the bottom of the basin. These issues again point to the importance of understanding local conditions and influent concentrations when selecting an appropriate drawdown rate.
Length-to-Width Ratio The purpose of specifying a length-to-width (L/W) ratio is to improve hydraulic efficiency by reducing short-circuiting and dead zones. With respect to wet basins, hydraulic efficiency is defined as how well fresh stormwater exchanges with older water in the basin. Table 1 indicates a large range in design values. Specifying an L/W ratio for extended detention basins is likely unnecessary as these basins operate essentially as fill-and-draw systems. The constricted outlet tends to force incoming stormwater into all areas of the basin. A comparison of six extended detention basins in which the L/W ratio varied from 3 to 10 found no difference in performance (Caltrans 2004). Whether performance is affected by ratios in the range of 1 to 3 is unknown.
While a high L/W ratio may maximize hydraulic efficiency, it may also create a geometry that is difficult to fit into many developments. Furthermore, excessively narrow basins may induce the resuspension of sediments because of greater throughput water velocities. The experience from wastewater treatment is to have substantial L/W ratios, on the order 10 to 20 or greater depending on the unit operation (e.g., sedimentation, chlorination). Such high L/W ratios are not necessary in stormwater systems because the rate of inflow is generally lower relative to the basin volume than in wastewater systems.
One study of the relationship between the L/W ratio and hydraulic efficiency suggests a ratio above 2 provides little additional benefit. Figure 4 is an interpretation of Walker (1998). It shows the effect of increasing the L/W ratio on hydraulic efficiency. Figure 4 suggests modest benefit of incremental increases of the ratio above 2. It also suggests the appropriate L/W ratio likely differs with the size of the basin as defined by the Vb/Vr value. For basins with large Vb/Vr values a L/W ratio of 2 is likely satisfactory. If the Vb/Vr of wet basins is on the order of 1 as recommended in this article, the L/W ratio should be greater, perhaps 3 to 4. The storm intensity common to the region may also be relevant to this decision. In the Southeast, with high-intensity storms, the incremental benefit of increasing the L/W ratio is likely greater than the Pacific Northwest with its mild storms. This difference is suggested by contrasting the results of Persson et al. (1999) and Persson (2000) to Walker (1996). In the former studies significant benefit was found up to a ratio of 5. However, these studies were conducted at higher flow rates.
Poor hydraulic efficiency has been found not to be an issue with poorly configured large wet basins with Vb/Vr values on the order of 10 (Pitt 2000). This is not likely the case with smaller Vb/Vr ratios. Baffles were added to a pond with a Vb/Vr of about 1.5, increasing the L/W from 1.5 to 4.5 (Mathews et al. 1997). A significant improvement in hydraulic efficiency was observed.
An L/W ratio of 1 is likely valid for small wet basins if an extended detention layer is included on top of the wet pool. Restricting the outlet causes stormwater to "back into" areas that would otherwise be dead zones such as corners or areas of dense vegetation. The designer may be allowed the flexibility of using either a wet pool only with a large L/W ratio, or a combination extended detention/wet pool with a smaller L/W ratio to meet site constraints. For pure extended detention basins a L/W ratio of 1 is likely sufficient.
As with the L/W ratio, the benefits of an extended detention layer may differ with the climatic region. The concept may be most beneficial in the Southeast with its short, high-intensity storms and high rates of inflow. In contrast, the concept may be of less benefit on the West Coast where average storm intensities are much less.
The above discussion does not take into consideration the potential adverse effects of thermal or density (from deicing salts) stratification on hydraulic and therefore performance efficiency. The effect of stratification is poorly understood. Thermal stratification reduced the detention time of a small wet basin by half (Timmins et al. 1999). The significance of stratification to performance is unknown, as is the effect of the energy of incoming storms to temporarily destratify the basin. One study concluded that summer storms abetted stratification due to the incoming stormwater being higher in temperature than the pond water (McBean and Burn 1983).
Stratification should not reduce performance during storms. As established above, performance is a function of hydraulic loading rate, which remains the same irrespective of stratification. However, stratification reduces the exchange of fresh stormwater with water present in the basin. As a consequence, stratification in effect reduces performance between storms with respect to fine sediments and dissolved pollutants. Increasing the L/W ratio might decrease the potential for stratification by increasing flow velocities within the basin. Subsurface discharge to the lower half of the wet pool and/or entrance baffles might have a greater mitigating effect than a large L/W. A final note of interest is that in the study cited above, thermal stratification occurred in the top 10 inches of the wet pool. This suggests that the common recommendation (Table 1) to limit the maximum depth to avoid stratification may not be effective. More studies on this subject are needed.
Multiple Ponds or Cells Several manuals suggest incorporating multiple cells, either as separate basins in series or by separation of one basin into two or more cells with a berm extending above the water surface. Multiple cells can significantly increase space requirements. The purported benefit is more effective treatment. However, there are no data substantiating this claim. Some conceptual designs like that shown in Figure 5 may result in poorer performance than one basin of equal volume. While aesthetically pleasing, multi-ponds with irregular shapes increase short-circuiting and dead zones. This decreases hydraulic efficiency and, as a consequence, performance efficiency. An extended detention layer may minimize this effect.
Vegetation Coverage
For wetponds, a shallow safety bench of 10 to 15 feet in width is commonly recommended. The benches become covered by emergent vegetation. For wetlands, manuals usually specify a depth regime that results in vegetation coverage on the order 60% - 85% of the surface area. Concepts of vegetation configurations are frequently included in manuals (Figure 5).
Depending on its degree and location, emergent vegetation may enhance or degrade hydraulic efficiency and therefore performance efficiency. If the basin is configured so as to produce an open area down the middle of the basin, hydraulic efficiency and therefore performance is reduced. Water, finding the path of least resistance, moves through the center of the basin, with little exchange with "old" water in the densely vegetated areas along the sides of the basin (Persson et al. 1999). In shallow marsh wetlands, low-flow channels may have a similar effect. As the wetland ages, a pattern of uneven plant densities may create channels of less resistance, decreasing hydraulic efficiency.
A more appropriate configuration might be open fore- and after bays with a shallow intermediate marsh area and no low-flow channel. To use less space, the intermediate area could be of wetpond depth. The interface between the central area and the fore- or after bays can be a berm, but with the top below the water and covered by emergent vegetation. This berm configuration could more evenly spread incoming water across the lateral cross-section of the pond. Alternatively, as previously noted, an extended detention layer might compensate for the adverse effects of fringe vegetation on hydraulic efficiency.
Sediment Storage
The common criterion is to either add 1 foot of depth or increase the basin volume by 20% above that calculated for performance. The added cost is not trivial. However, once construction of the development is complete, the accumulation rate of sediment is modest, on the order of 0.25 to 0.5 in/yr. Furthermore, most of the incoming sediment settles in the forebay. It is suggested that this requirement be eliminated, particularly if a forebay is inherent to the design. Furthermore, a more cost-effective alternative to the earthen forebay might be a large standard manhole, a small manufactured wet vault (e.g., Stormceptor), or a vortex separator. Even if the initial cost is greater, the life-cycle cost might be lower given the greater ease of maintenance of these devices.
Final Observations
The discussion suggests that wet basins and extended detention basins are not separate types but opposite ends of the same type. A continuum exists between the two extremes, depending on the climate of the particular region and the design volume. The discussion further suggests that extended detention basins should perform as well as wet basins, although they must be larger to achieve similar performance with respect to the removal of sediment and particulate pollutants. As previously noted, particles as small as 5 to 10 microns should be readily removed at current drawdown rates. However, performance studies suggest otherwise, as implied in Figure 3. TSS in the effluent of wetponds are typically on the order of 10 to 20 mg/L, whereas they are on the order of 20 to 40 mg/L for extended detention basins. This discrepancy is due in part to wetponds having volumes equal to or greater than those of extended detention basins (Table 1). It may also be due in part to the likely inability to induce the coagulation of clay even with a specified drawdown time for small storms.
Design aspects other than volume and drawdown time might contribute to a less-than-expected performance for extended detention basins. The most likely reasons are adverse inlet and outlet conditions. Resuspension of previously deposited sediment and erosion may occur in the inlet area during the infrequent high-intensity storms if energy dissipation is insufficient. Resuspended clays might have insufficient time to re-coagulate and as a result exit during these events. The more likely candidate for subpar performance is the outlet structure. In a water or wastewater basin, exiting water is gathered along an effluent weir that extends the width of the basin. This minimizes approach velocities to the weir. By contrast, the outlet structures of extended detention basins are very small relative to their end-cross sections. As a consequence, stormwater "rushes" toward one relatively small point in the basin. Approach velocities near the outlet are excessive relative to the settling velocity of silts and clays, and even fine sand. The effective HLR in the vicinity of the outlet is substantially greater than the nominal HLR of the basin. Sediments that should be removed based on the nominal HLR, or drawdown rate, are lost through the outlet. This condition is exacerbated if the bottom around the outlet is bare, allowing for erosion at the high approach velocities. While wet basins also have narrow outlets, the effect is less severe. With wet basins the water exiting each storm is "old" water that is relatively clear of fine sediments, given the retention time between storms.
One or a combination of design elements may mitigate the effect of small outlet structures. Several outlet risers tied to an exit manifold may help. However, the issue of clogging very small orifices arises. A shallow wet pool or micropool at the outlet may counter the effect of excessive approach velocities by preventing scouring of accumulated sediments at the bottom of the basin in the vicinity of the riser structure (Fennessey and Jarrett 1997). Some have found that withdrawal at the surface improves performance (Ward et al. 1979; Millen et al. 1997) whereas others have not (Keblin et al. 1998). Spacing, size, and the location of perforations on the riser may affect performance (Ward et al. 1979). However, another study found no significant difference in performance between a single bottom orifice and a perforated vertical riser (Fennessey and Jarrett 1997). It is possible that while the riser configuration reduces approach velocities, the reduction may not be sufficient to significantly affect performance. Certainly, however, a riser structure with a series of vertically placed holes should reduce erosion and/or resuspension in the immediate vicinity of the outlet. A cone of gravel around the bottom of the outlet structure has been found to significantly improve performance (Engle and Jarrett 1995). More studies on this issue are needed. Basins sized as suggested here should be able to meet the general criterion of 80% removal of TSS, a specification of most manuals. They should also be able to meet the specification of 40% removal of the total phosphorus (TP) inasmuch as 75% - 90% of the phosphorus is typically bound to the sediments. Meeting the goal of 40% removal of nitrogen is less certain given that it is more soluble than phosphorus. These questions are considered in the fourth article of this series.
It is common to specify a minimum drainage area, under the thesis that a base flow of water into the basin is needed to keep a permanent pool. This limits the use of wet basins for larger areas, on the order of at least 10 to 25 acres depending on the region. The concern is that the wet basin will not fully function without water. This is not the case. There is no need for the basin to "work" between storms. Pollutants migrate to the bottom as water evaporates and into the soil with infiltration. Allowing the pond to dry may reduce mosquito problems. Infiltration is good for the hydrologic cycle. However, drying may solubilize some pollutants, which might be lost in subsequent storms. If the removal target is solely TSS and attached pollutants, this is no concern. Even if dissolved pollutants are of concern, the pond or wetland fills with the next storm, at which time the solubilized pollutants likely renew attachment to inorganic and organic sediments. Constructed wetlands can withstand extended periods of dryness without harm to the vegetation. Specifications as to the maximum length of dryness to avoid this problem may be prudent and will vary with the climatic region and vegetation species. It is also important to recognize that a constant base flow can reduce the performance with some pollutants. One study found a negative net removal of phosphorus (Oberts 1997). Removal during storms was offset by loss in base flows during dry periods.
The concepts discussed in this article are also relevant to manufactured basins such as Stormceptor, BaySaver, EcoStorm, and Stormvault. Regardless of how they may be configured, they are essentially wet vaults.
Summary
The data suggest wet basins can be smaller than currently specified. They also can be smaller than extended detention basins if the sole objective is the reduction of sediment and attached pollutants such as metals, pesticides, and nutrients. In most situations this "basic" level of treatment is likely sufficient. A dead storage volume equal to the runoff of the mean annual event‹that is, a Vb/Vr of 1 to 1.5‹is likely adequate for wet basins. A larger volume may be necessary where significant removal of dissolved pollutants is also desired. This aspect is considered in the fourth article of this series.
To achieve reasonably equivalent performance, extended detention basins must be larger than wet basins with a Vb/Vr on the order of 2 to 2.5, depending on the region. The specification of volume must also take into consideration the specified drawdown time at brimful and the interevent time for the locality. A longer specified drawdown time and shorter interevent time requires a larger basin volume to achieve the same management objective with respect to the aggregate volume of water treated over time. The effect of infiltration and evaporation could be included in this analysis.
Drawdown rate rather than drawdown time is the correct design criterion for extended detention basins. In some regions, in particular the West Coast, recognition of drawdown rate as the correct criterion results in more cost-effective sizing. Current data suggest the specification for the average drawdown rate from brimful be on the order of 0.15 ft/hr. Although there is considerable uncertainty with this specification, the same uncertainty exists with the current specification for drawdown time. Regardless, even if changing from drawdown time to rate has little effect on basin volumes for much of the nation, the use of drawdown rate will ensure that field studies gather the appropriate information that has been lacking in most previous studies.
Drawdown time may be relevant to the extent that clay is a significant fraction of the incoming sediment. Local studies are needed to ascertain the relevance of clay as are more studies on the relationship between time, coagulation, and settling. The data suggest, however, that the issue of clay might not be the most significant factor in what appears to be less-than-expected performance of dry basins.
There may be an inherent limitation of extended detention basins to perform as well as wet basins irrespective of the drawdown rate or time. Design aspects other than volume and drawdown time must be considered, in particular the outlet structure. Modifications to current outlet design specifications are needed to reduce approaching velocities and the potential for resuspension and/or erosion in the vicinity of the outlet. The limitations of these basins can be mitigated through the inclusion of a wet pool. However, at some point the question arises as to what is being designed: an extended detention basin with a wet pool or a wet basin with an extended detention layer. It follows that at some point the relevance of an extended detention layer becomes problematic.
The length-to-width ratio is an important design parameter for wet basins, but it is probably not particularly relevant to extended detention basins. A ratio on the order of 2 to 4 is likely sufficient for wet basins. The appropriate ratio depends on the selected Vb/Vr and general rainfall intensity. A ratio of 1 is likely sufficient for extended detention basins. Oddly configured basins or basins with varying bathymetry adversely affect hydraulic efficiency and therefore performance efficiency. The adverse effect of such configurations can be reduced by increasing the size of the system and/or by including an extended detention layer. However, as these relationships are little understood, it is advisable to carry out the appropriate studies before imposing such requirements.
Care must be taken with emergent vegetation. Depending on location and density, emergent vegetation can be counterproductive. Vegetation along the sides of a wet basin, with open water down the center, reduces hydraulic efficiency. This can be countered by lateral berms with emergent vegetation and/or an extended detention layer. The significance of this observation is greater in regions with high-intensity storms such as the Southeast.
Additional laboratory and field studies are needed to better define the significance of the observations and recommendations made in this article. Being aware of the relevant design criteria and how their selection may affect performance leads to studies in which the most relevant characteristics of performance are evaluated such as hydraulic efficiency, water level fluctuations, and the size distributions of incoming and outgoing sediments.