A Holistic Watershed Approach for CSO Separation and Removal

Under the Clean Water Act, the federal government has required cities to reduce and eliminate combined sewer overflows (CSOs). Cities are required to comply with those requirements to avoid penalization through fines and other legal actions. The Metropolitan Sewer District of Greater Cincinnati (MSDGC) is under a consent decree to respond to CSOs.

The Ludlow Run watershed assessment was conducted in an effort to eliminate combined sewer discharges from seven CSOs located within the watershed. An innovative screening tool was developed using the existing Cincinnati Area Geographic Information System (CAGIS) database, to evaluate stormwater drainage and determine suitable locations for stormwater best management practices (BMPs) on a parcel-by-parcel basis for the entire 2.5-square-mile watershed. The main focus of this screening tool was development of a BMP suitability index, which assessed factors such as hydraulic soil group, slope, and land use.

Stormwater management alternatives were assessed using the stormwater management modeling software XPSWMM to model watershed hydrology at the sub-basin scale, including hydraulics of open channels and enclosed systems of stormwater and wastewater flows. The hydrologic component of XPSWMM divided the Ludlow Run watershed into smaller sub-basins, using topographic data provided by MSDGC. Hydrologic processes, including infiltration, evaporation, ponding, and ground-surface water exchanges were included in the model.

The XPSWMM model included the Ludlow Run main stem and all major tributaries. The open channel network, reach lengths, and cross sections were derived from a 3D surface model. Boundary conditions, roughness coefficients, channel crossings, and obstructions were assessed during on-the-ground field reconnaissance of the project site.

Evaluation of stormwater infrastructure to intercept and convey stormwater runoff and restore natural drainage conditions included a combination of grey and green design approaches, including separate sanitary and stormwater pipes, subsurface stormwater storage, biofiltration, vegetative bump-outs, curbside planters, pervious pavers, bioretention wetlands, and riparian corridor restoration. A detailed alternatives analysis assessed viability, benefits, and potential cost savings to MSDGC’s ratepayers. The alternatives described in this article are the recommended solution to eliminate CSO 24 and the contributing nested CSOs.

Background

Through negotiations with USEPA, the US Department of Justice, and the State of Ohio, MSDGC drafted a consent decree that formed the response program to resolve its overflows. The global consent decree outlines MSDGC’s plan of action to reduce and eliminate CSOs, enhance the environment, and pay fines for past overflow events. MSDGC is now utilizing funding provided by the decree to examine alternatives for overflow removal. In an effort to enhance the environment and eliminate CSOs, MSDGC is reviewing the effectiveness of ecologically beneficial (or green) BMPs compared with traditional wastewater infrastructure (grey infrastructure).

The Ludlow Run watershed assessment was conducted in an effort to eliminate combined sewer discharges from seven CSOs located within the watershed (Figure 1). During periods of heavy rainfall or snowmelt, mixed stormwater and wastewater exceed the capacity of the existing combined sewer system and discharge directly to Ludlow Run, a tributary to the Lower Mill Creek in Cincinnati. CSO 24 is the combined sewer overflow location and the endpoint of Ludlow Run that discharges into the Mill Creek. A map of CSO 24 and its six nested CSOs (109, 110, 111, 112, 151, and 162) is shown in Figure 2.

BMP Site Suitability Index

Slope is an important parameter in BMP design. The performance and implementation of stormwater detention, retention, and conveyance systems are limited by steep slopes. To evaluate locations based upon slope, and to identify contiguous areas of shallow slopes, localized gradients were needed for the entire watershed. ESRI’s ArcView 9.3 was used to calculate localized slopes at a resolution of 5 feet by 5 feet. Contour lines were converted into a topographic surface. Then, using standard ArcView commands, the contour polylines were converted into a surface raster with a resolution of 5 feet by 5 feet. This divided up the entire sewershed into 5-foot by 5-foot squares or cells and assigned each cell an averaged elevation based on the original contour lines. Localized slope was calculated for each cell and then assigned a point value based upon the calculated slope. Point values ranged from 0 to 300 points (>20% is equal to 0 points, whereas 0% to 5% is equal to 300 points).

The hydraulic soil group (HSG) rating is a measure of a soil’s runoff potential. The rating is based on the soil layer with the lowest saturated hydraulic conductivity and the shortest depth to an impermeable layer. There are four HSG ratings, Groups A to D. At one end of the spectrum, Group A soils have a low potential to create surface runoff when the soil is thoroughly wet. These are typically soils with less than 10% clay and greater than 90% sand and gravel. At the other end of the spectrum, Group D soils have a high potential to create surface runoff when the soil is thoroughly wet. These soils are typically greater than 40% clay and less than 50% sand. Because the soil data were not available in CAGIS, each parcel was manually assigned a rating based upon the predominant HSG from the Web Soil Survey. Each soil type was assigned a point value of 100 (Group D) to 300 points (Group B).

In evaluating land use, the range of existing land uses was divided into three categories: open space, woods, and “other” developed land uses. Due to the cost-prohibitive nature of acquiring and redeveloping a residential or commercial property, preference was given to parcels that were either open space or forested. Each parcel was assigned a land use manually, based on review of the latest available aerial photographs. Land uses were assigned point values ranging from 0 for developed property to 500 for open, undeveloped lands.

After the evaluation criteria were selected and applied to the entire watershed, the criteria were assigned point values, with higher points indicating a higher suitability for BMP placement. A color gradation map was used to interpret the range of point values, with green parcels indicating sites most suitable for regional BMPs and red indicating sites that are unsuitable for BMPs (Figure 3). Land use received the highest weighting, because this criteria has the greatest impact on both cost and construction feasibility. The scores were summed for each cell and ranged from a minimum of 100 points (other land use, HSG D, and >20% slope) to a maximum of 1,100 points (open space, HSG B, and 0% to 5% slope). Areas shown in green are the most suitable locations for BMP placement, while those shown in red are the least favorable. This process allowed for the identification of 10 potential BMP locations (areas A through J as shown in Figure 4) that were to be investigated further in the field.

Field Verification of Site Suitability

Locations C, E, F, G, H, and I were deemed suitable for green and/or grey BMPs. The natural drainage paths of location C flow to Ludlow Run, making this a good location to collect and convey the stormwater from the rest of the sewershed. The low-lying, flat areas of location C show signs of ponding water during storm events, suggesting a suitable location for constructed wetlands. The floodplain along Ludlow Run at location E was cleared and leveled to allow construction of the new combined sewer line. This allows for the construction of future floodplain modifications, such as floodplain wetlands. A wetland already exists near the southeast end of location F, which flows to a tributary to Ludlow Run. The northwest end of location F (just upstream of CSO 110) is flat and suitable for an additional storage BMP that could intercept the street runoff. Location G was determined to be suitable for conveyance BMPs, but not suitable for storage BMPs. The northeast (downstream) end of location G was too steep to accommodate a storage BMP. However, the topography of the site was suitable for conveying the stormwater from the streets to the tributary and ultimately to Ludlow Run.

Location H was determined to be suitable for floodplain storage BMPs. The floodplain in this area has signs of frequent inundation, which suggests that flows would often reach the floodplain and thus floodplain storage BMPs. The southern section of location I was determined to be suitable for green and grey BMPs, while the northern section of location I was determined to be suitable only for floodplain storage. In the southern section, the topography naturally drains from the CSO 162 sewershed down to location I, providing ideal conditions to convey and store stormwater. Due to the pipe configuration and topography, it was infeasible to route stormwater under gravity flow to the northern section of location I.

Development of Stormwater BMP Palette

Vegetated Bump-out. A vegetated bump-out (or curb extension) is a vegetated matrix that consists of two layers: gravel at the bottom and growing media at the top. The bump-out is placed in the direct flow path of the existing curb and gutter designed to intercept gutter flow. The bump-outs fill with water, capturing the first pulse of the storm and removing volume through infiltration and eventually evapotranspiration. The bump-outs overflow back into the gutter and into the existing grated inlets. Bump-outs can be installed in place of an existing parking lane area (Figure 5) or extend into an existing tree lawn for greater surface area and storage volume. The length and slope of each bump-out varied, based on the available street length to install the bump-out and the slope across it. To account for each variation, the model was run for possible slope and length combinations.

Pervious Pavers. Another green technology suitable for capturing and storing street runoff is pervious paver parking lanes. Pervious parking replaces the existing asphalt surface with porous media with a high infiltration rate, allowing the stormwater runoff flowing over it to infiltrate rather than flow directly to a grated inlet (Figure 6). The pervious parking cross section consists of three layers. The bottom layer is a gravel storage layer. Above that is a compacted leveling layer required to create a smooth driving surface. Lastly, the top layer is a set of interlocking pervious pavers, similar in appearance to cobblestone. Pervious pavers are generally not suited to high-speed or high-traffic areas and therefore, were only used as a replacement in the low-traffic parking lanes.

Wetlands. Stormwater wetlands are regional detention BMPs, designed to provide extended detention of stormwater and to maximize pollution attenuation through microbial activity, plant uptake, retention, settling, and adsorption (Figure 7). In addition to stormwater storage, wetlands also provide treatment value and provide wildlife habitat. For conceptual design purposes, the wetlands proposed in each design alternative were assumed to have an average depth of 2 feet and a drawdown time between 48 and 72 hours.

Step Pools. Step pools are open channel conveyances for steep terrain (Figure 8). These systems rely on an alternating sequence of rock riffles, or drop structures, and constructed pools for energy dissipation. The step pool storm conveyance method was used to conceptually size the step pools for the alternatives identified in this assessment. Each riffle is 10 feet long and has a 5-foot elevation change from upstream to downstream end. Each pool is 10 feet long and 2 feet deep. The riffles and pools are combined in a riffle-pool-pool-pool sequence. This combination yields a 5-foot drop every 40 feet, or a 12.5% slope.

Underground Storage Vault. Underground storage vaults, consisting of buried large-diameter pipes, provide watershed-scale storage, reducing peak and total flows entering the combined sewer system or to streams and rivers. Pipes can be placed in parallel or in series with each other to provide the required storage volume. The pipes can be designed to have the same 48- to 72-hour drawdown time as extended detention wetlands. This extended drawdown is critical to reducing erosion at the outfall and to prevent downstream flooding in Ludlow Run and its tributaries.

Performance Evaluation

The hydrologic component of the XPSWMM model divided the Ludlow Run Watershed into smaller sub-basins, using topographic data provided by MSDGC. Existing land-use data from aerial photographs and CAGIS, coupled with USDA Natural Resource Conservation Service soils data, were used to determine runoff characteristics of each sub-basin. These sub-basin characteristics, stormwater flow paths, and published rainfall intensity/duration data were incorporated into the XPSWMM model. Hydrologic processes, including infiltration, evaporation, ponding, and ground-surface water exchanges, were included in the model. The XPSWMM model included the Ludlow Run main stem and all major tributaries. The open channel network, reach lengths, and cross sections were derived from a 3D surface model. Boundary conditions, roughness coefficients, channel crossings, and obstructions were assessed during on-the-ground field reconnaissance of the project site.

Evaluation of stormwater infrastructure to restore natural drainage conditions included a combination of grey and green design approaches, including separate sanitary and stormwater pipes, subsurface stormwater storage, biofiltration, vegetative bump-outs, curbside planters, pervious pavers, bioretention wetlands, and riparian corridor restoration.

Selected Alternative for Each CSO

CSO 151. The selected alternative for CSO 151 is designed to capture stormwater on the streets, before it reaches the combined sewer system, using vegetated bump-outs and pervious paver parking lanes. A total of 31 bump-outs fit logistically into the current street layout, placed only in existing parking lanes (and avoiding driveways or crosswalks). An additional 16,000 feet of pervious paver parking lanes would be installed, replacing the existing asphalt-covered parking areas. The addition of green infrastructure would be coupled with the reliance on existing stormwater-only lines. In their present configuration, these lines intercept nearly 50% of the CSO 151 watershed but flow directly to the combined sewer system. The strategic use of additional piping could effectively separate the stormwater lines from the combined system, routing stormwater to location C, where it would discharge into an extended detention basin or constructed wetland.

CSOs 109, 110, and 112. With an approach similar to CSO 151, the selected alternatives for CSOs 109, 110, and 112 rely on the ability to capture and detain stormwater on the streets before it reaches the grated inlets leading to the combined sewer system. For CSO 109, this could be accomplished using a combination of seven bump-outs and 2,400 feet of pervious pavers. For CSO 110, only four bump-outs and 1,490 feet of pervious paver parking lanes could be used, due to site constraints including steep topography and narrow streets. CSO 112 would employ nine bump-outs and 2,690 feet of pervious paver parking lanes. Taking this approach one step further, a hybrid alternative could completely separate the stormwater from the combined sewer system by disconnecting existing grated inlets from the combined system and connecting them to new stormwater-only lines. These lines would then discharge to an open channel conveyance, transporting stormwater to Ludlow Run or its tributaries.

CSO 111 and 162. For other sewersheds, the solution to CSO separation is not nearly as complex. The preferred alternative for CSO 111 is a simple disconnect of residential downspouts, routing the flow of stormwater into vegetated swales. For CSO 162, complete separation can be achieved by closing existing inlets to the combined sewer system and discharging into a step-pool conveyance BMP. The step pools would flow into a wetland located in the southern section of location I.

CSO 24. The proposed separation alternatives for the CSOs described above rely on the removal of the largest combined sewer conveyance at the lowermost point in the watershed, CSO 24. This CSO location is the endpoint of Ludlow Run that discharges an estimated 318 million gallons into the Mill Creek every year. To completely separate CSO 24 (and the nested CSOs described above), the installation of a separate sanitary line would be required, thereby converting the existing Ludlow Run conveyance tunnel into a stormwater-only line (Figure 10). The CSO 24 overflow structure would then be effectively converted into a natural confluence with the Mill Creek, eliminating all CSO overflow events for a typical year.

Conclusion
The CSO 24 Separation Project illustrates the benefits of using a holistic watershed approach to CSO separation projects. By identifying existing constraints and opportunities early in the planning process, we were able to maximize schedule and budget to streamline the process. While there is an extensive menu of stormwater BMPs to choose from, each with particular requirements and limitations to function properly, determining a palette of BMPs suitable to a particular region of operation will focus design efforts. Equally as important is the process of field verification, getting “boots on the ground” to verify location and suitability of conventional and green stormwater BMPs. Regardless of the location, watershed, existing infrastructure, or proposed BMPs, a rigorous performance evaluation using XPSWMM or similar modeling software is an essential component to a watershed-scale management plan. By evaluating and modeling multiple alternatives, we were able to provide workable solutions to demonstrate separation and eliminate 328 million gallons of overflow to Mill Creek from the Ludlow Run watershed.