Treatment Trains

Don’t get run over.

The term treatment train has become ensconced in the lexicon of stormwater treatment terminology since first introduced (Horner and Skupien 1994). There are several variants. Two are presented below.

• Concept 1: A set of source control best management practices (BMPs), possibly followed by a treatment device
• Concept 2: A series of separate treatment devices or “boxes”

Concept 1 describes a menu of source control BMPs—cleaning of catch basin sumps and sweeping—followed by, for example, a wet vault. Concept 2 is a series of treatment units, which we will call unit operations.

The focus of this article is on the second concept. Caution is advised regarding implementation of the second concept. A general rule to follow: Do not use two unit operations that serve the same function, as described later in this article.

Figure 1 shows an example of a common treatment train: a wet pond followed by a shallow marsh wetland. The relevant question is whether the combination shown in Figure 1 provides more effective treatment than a single wet pond or a single constructed wetland. And, even if the combination is more effective, is the incremental improvement worth the marginal increase in cost? Other not-uncommon configurations are listed below:

• Wet pond + wetland
• Forebay + silt trap + flood detention basin + marsh + swale
• Swale + “prairie” wetland + marsh wetland
• Oil/grit separator + sand filter
• Swale + wet pond
• Oil/grit separator followed by a dry well
• Extended detention basin + sand filter
• Vegetated strip + infiltration trench

The last two examples are not normally viewed as treatment trains, but the reason for their inclusion will become apparent.

Figure 1. Wet Pond - Constructed Wetland Treatment Train

Observed Performance of Treatment Trains
Unfortunately, field studies that have evaluated the performance of treatment train sequences have typically limited the sampling to the influent to the first unit and the effluent from the final unit. Infrequently are samples taken at the intermediate points between the treatment units. From such studies, it is not possible to ascertain the incremental contribution of each treatment unit.

There are exceptions. Table 1 presents the results from one field study where sampling occurred at the intermediate points (TRCA 2002). The particular system consisted of three separate basins in sequence: sediment basin, large wet pond, and relatively small wetland. The surface area of the wetland was about 20% of the wet pond. Table 1 suggests the wetland provided no additional performance benefit except phosphorus removal. The wetland decreased the performance for ammonia. The fact that performance can be adversely affected by the addition of a treatment unit has been overlooked.

A few other available studies provide additional insight. A study of the sequence of wet pond plus grass swale found the median total suspended solids (TSS) concentration increased from 2 to 14 milligrams per liter (Colwell 2001) as stormwater passed through the swale. Phosphorus was found to exit a wet pond/wetland sequence during base flow (Oberts 1999), likely by desorption of dissolved phosphorus. The effect was no net removal of phosphorus. A study of a wet pond/wetland system found the concentrations of several metals, nitrogen, and phosphorus increased as the stormwater passed through the wetland (Gain 1996). Performance based on concentration can be misleading, because concentrations may be affected by evaporation or direct entry of groundwater into the bottom of the facility. Nonetheless, the limited data suggest caution.

Unit Processes and Unit Operations
The author has proposed (Minton 2002, 2005) that more rational decisions on treatment system selection and design would occur if stormwater engineers adopted a concept long established in chemical engineering, potable water supply engineering, and wastewater engineering: unit processes and unit operations (Rich 1961, 1963). Definitions of the two terms have been modified to make them relevant to stormwater treatment (Minton 2002). As defined by Minton (2002) for stormwater treatment, unit process is a mechanism of pollutant removal; unit operation is a structure in which one or more processes occur.

Examples of unit processes are sedimentation, adsorption, and filtration. Examples of unit operations include a wet pond, a sand filter, and a vortex separator. Several unit processes may occur in a unit operation. Two or more unit operations constitute a treatment train (the author prefers to use the word system). The concept is illustrated in Figure 2, which shows a filter system with two unit operations: a wet vault and a filter. Several unit processes occur within each unit operation: sedimentation and flotation in the wet vault and sedimentation and filtration in the filter, as well as sorption and precipitation depending on the media and the characteristics of the stormwater. The wet vault pretreats the stormwater, reducing the maintenance frequency for the filter. It removes the coarser sediments such as sands and some silt. It could be excluded, and should be if the stormwater is expected to have a relatively low suspended solids concentration. But for typical stormwater, exclusion of pretreatment will likely result in an unacceptable frequency of cleaning the filter. Although not thought of as such, Figure 2 is in fact a treatment train.

Figure 2. Example of Unit Operations and Unit Processes

Table 2 presents the likely unit processes in various unit operations. Table 2 summarizes unit processes that occur in each treatment technology, divided into three categories: physical, chemical, and biological. Significant unit processes are shown in italics. With the lack of data, opinions differ with some designations: for example, the role of plants in wetlands and swales. Significance varies with the technology. For example, the demarcation between wet ponds and wetlands with respect to vegetation cover is not distinct. The significance of some processes in some unit operations is not well understood at this time.

Consider Stormwater Quality
Understanding stormwater quality leads to insights into the appropriateness of treatment trains: when they might be applied and limitations of their use. Stormwater pollutants are commonly divided into two groups: suspended solids or particulate pollutants, and dissolved pollutants. The division point is defined by the size of the openings of the laboratory filter, commonly 0.45 micron. The material retained on the laboratory filter is defined as suspended. What passes through the filter is defined as dissolved.

Suspended solids come in various sizes. Therefore, the group is further characterized by its particle size distribution (PSD). The PSD can be translated into a settling velocity distribution, through calculations or direct measurement (Minton 2005). The settling velocity distribution is directly germane to the performance of basins and filters. Generally, most pollutants tend to be associated with the finer suspended solids, clays, and silts, although there has recently arisen some dispute over this long-held view (Sansalone et al. 1998).

Most pollutants readily sorb to suspended solids in the stormwater before they reach the treatment device. Some notable exceptions are copper, cadmium, zinc, some pesticides, and nitrogen and phosphorus, but to a lesser extent than the cited metals. While classified as dissolved, pollutants that pass through the laboratory filter are not entirely in a free form and therefore are not necessarily readily available for sorption to filter media or soil in the bottom of a pond. The pollutant sorbs to very fine, submicron colloids that pass through the laboratory filter. The effect differs with the pollutant and water chemistry (Minton 2005). Furthermore, most pollutants are at very low concentrations in untreated stormwater, making substantial removal difficult. As the primary mechanism to suspended solids or sorptive media in the case of soils or filter media is adsorption, desorption can occur as well. A laboratory study observed desorption of dissolved metals from filter media at low influent metals concentrations (Johnson et al. 2003). The metals were likely original to the media. A roadside strip in which compost was mixed into the soil resulted in an increase in metal concentrations (Yonge 2000).

Consider the Size of Each Unit Operation
Sizing complicates the discussion. Recommendations for sizes of individual unit operations in various configurations of treatment trains vary from inconsistent to nonexistent. One view is that two full-size units should be used; that is, each unit could stand alone, and each can meet a basic performance goal, 80% removal of TSS, with the intended purpose of the second box to provide further treatment for particular pollutants such as phosphorus, nitrogen, or metals (WSDOE 2001). More commonly, only one of the units may be “full size,” as is apparently the case for the system represented in Table 1.

In the case of Figure 1, we might consider two extremes. Typically, the required volume of a basin is defined as equal to the water-quality or treatment volume. This volume is equal to a specified storm runoff depth, such as 1 inch (2.5 centimeters) times the drainage area and runoff coefficient. One extreme is to have the volume of the wet pond and wetland each equal to the water-quality volume—in effect, two full-size unit operations. The opposite extreme is to have the combined volume of the two equal to the water-quality volume. For the first case, the incremental benefit of the constructed wetland is doubtful, if presumably substantial performance is achieved. As noted previously, both unit operations do essentially the same thing: remove suspended solids and both particulate and dissolved pollutants. It is questionable that the wetland will provide noticeable improvement of the effluent from the wet pond. A possible exception is dissolved phosphorus, given the importance of plant growth in long-term sequestering (Minton 2005). The observation is suggested by the results in Table 1. This illustrates the importance of identifying the specific pollutant of interest.

For the second case, how the water-quality volume is distributed between the two unit operations is a consideration. The value of the distribution implied in Figure 1 is problematic. A better distribution is illustrated in Figure 3, where the first unit operation is the forebay: its function to capture the coarser sediments, which likely predominate. The objective is to ease maintenance. As shown in Figure 3, the second unit operation is a wet pond, but it could be a wetland as in Figure 1. A final observation on Figure 1: This configuration could possibly reduce overall performance due to the creation of many dead zones in the corners of each unit. Constricting the outlet of each or both may reduce the effect, but this adds to the required volume. However, there is no evidence that it works.

Figure 3. Water-Quality Volume Distribution

A similar concept—a first unit to reduce maintenance costs—is used with sand filters, as previously discussed and shown in Figure 2. One system known as the multi-chamber treatment train consists of several elements, but primarily a presettling chamber preceding a mixed-media filter (Pitt et al. 1997).

Consider the Effect of the First Unit Operation
Consider Figure 1. The wet pond, the first unit operation, removes coarse suspended solids (sand) and much of the fine material (silts and some clay). Hence, the PSD of the stormwater has been altered significantly. As the wetland is dealing with very fine material, its size, whether defined by surface area and/or water volume, cannot be reduced, even though preceded by the wet pond. To anthropomorphize, the clay and fine silts do not know they have entered a second unit operation where they are supposed to settle. Clay will be removed to the extent that the wetland retains some water between storms, allowing time for the clays to settle. However, this would also occur if the wetland was excluded and the volume of the wet pond was increased to include the volume of the wetland. (The wetland has a mechanism that is not significant in the wet pond. Colloidal sediments sorb to wetland plants. But the additional removal appears modest as suggested by the data in Table 1, and it may be offset by the loss of organic debris from decaying plants.)

A treatment train with two unit operations that both remove dissolved pollutants may find little removal in the second unit operation. In fact, the concentration of some dissolved pollutants may increase, as is the case with ammonia as shown in Table 1. With respect to Figure 1, the soil in the bottom of the wet pond will, if properly specified, effectively remove dissolved metals, pesticides, nitrogen, and possibly phosphorus. The plants in the wetland may increase the concentrations of some of these pollutants. For example, experience with wastewater wetlands indicates that plants generate a background concentration of nitrogen, on the order of 1 to 2 milligrams per liter (Minton 2005). The concentration of nitrogen in stormwater is commonly 1 to 2 milligrams per liter. If half of the nitrogen is removed in the wet pond, by settling and by nitrification-denitrification in the pond bottom, it may increase in the wetland through the cycling process of the plants. This may explain the increase in ammonia in Table 1.

Consider “Function”
The mistake with the treatment train in Figure 1 is that both unit operations serve the same function: removal of suspended solids and dissolved pollutants. The appropriate solution in this case is to have either the wet pond or the wetland, but not both. Alternatively, the size of the wet pond can be decreased, becoming a forebay, either in a separate basin or at the entry of the wetland. As noted previously, Figure 3 illustrates a better concept. In this case, even though the two unit operations remove suspended solids, the forebay serves to ease maintenance. Hence, it serves a different function: to reduce maintenance costs. Or it can be stated that the function of the wet pond/forebay is to remove coarse suspended solids, whereas the function of the wetland is to remove fine suspended and dissolved solids.

It should be noted that even if the wet pond/forebay is placed in the entrance area of the wetland, we should still view it as a separate unit operation. A vortex separator, for example, can serve the same function, possibly at lower cost. When it is viewed in these terms—as a unit operation serving a function—the engineer may realize there are more cost-effective approaches to providing the same function.

Other possible functions of individual unit operations are presented in Table 3. Reliability, defined as more consistent performance from storm to storm, may become a consideration with designated water-quality-limited water bodies for which total maximum daily loads have been determined.

Calculating Efficiency
Caution is offered regarding estimating efficiency of a treatment train, based on the observed efficiencies of each unit operation as a separate system (Horner and Skupien 1994; Minton 2005). In calculating the efficiency of a treatment train, we must recognize that the first unit significantly alters the characteristics of the entering stormwater.

Advertisement

Again consider Figure 1. Each unit operation operating separately might be expected to remove 80% of the suspended solids. However, placing the two in series will not except by happenstance produce a 96% reduction (80% plus 80% of the remaining 20%). The constructed wetland, the second unit operation, is treating stormwater with only fine suspended solids. It cannot be expected to remove 80% of the fine sediments, as a wetland is commonly sized. At least one manual recognizes this situation, specifying that when placed as the second unit, the efficiency of a unit operation is taken to be half were it the first unit (GDNR 2001).

Recommendations

• Stormwater engineering should adopt the concept of unit processes and unit operations.
• Apply the term treatment train to source control. For a series of treatment unit operations, use the term system.
• Follow the Golden Rule: Don’t place in a treatment system two unit operations that have the same function.
• Conversely, follow the second Golden Rule, which is to have a different function for each unit operation in the system.
• When considering the selection of a unit operation, the specific pollutant (for example, phosphorus) or pollutant type (for example, metals) to be removed should be identified, rather than thinking in terms of a general removal of multiple pollutants.
• Any two (or more) treatment elements in a system should be considered separate unit operations, whether in separate basins or separate compartments, or even if matched seamlessly, as in the case of a wet pool forebay in a larger pond.
• Recognize that including a second (or third) unit operation may provide little added benefit and may degrade performance.
• The additional expected benefit of an additional unit operation should be compared to the incremental cost of the added unit operation.
• Be careful when calculating efficiency of the treatment system. Recognize that a unit operation alters the characteristics of the stormwater that enters the next unit operation.

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.