Guidelines for Evaluating and Selecting Modifications to Existing Roadway Drainage Infrastructure to Improve Water Quality in Ultra-Urban Areas (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

83 This section presents guidance for planning-level sizing and design of volume-based and flow-based BMPs. A volume-based BMP is one whose performance is lim- ited by size and nature of its storage volume together with the drain time. Volume-based BMPs include most detention/ retention basins, underground vaults, and some bioretention installations. In this section, sizing guidance for volume-based BMPs is provided in the form of performance curves that relate the design variables (BMP storage volume and drain time) to performance metrics (runoff capture and sedimentation effi- ciency). The sizing guidance is applicable to a variety of surface detention basins or underground detention vaults. A flow-based BMP is one whose performance is limited by the flow rate passing through the BMP. Flow-based BMPs include media filters; filter strips; water quality inlets; and most small, proprietary systems that rely on hydrodynamic separation of particles. This section presents sizing guid- ance for generic non-proprietary media filtration systems in the form of performance curves that relate the surface area and detention volume to the runoff capture and media contact time. The BMP performance curves are based on continuous hydrologic simulation analyses for each of the 15 climate divisions specified by Driscoll et al. (1989). An accompany- ing spreadsheet tool provides users with planning-level sizing estimates for detention and media filtration BMPs in each of the 15 rain zones and allows users to explore tradeoffs between BMP design criteria and performance metrics. 6.1 BMP Sizing and Design Analysis Approach 6.1.1 Overview Design Storm Analysis: Event-based analyses are the most common approach for BMP sizing and design. Municipali- ties and DOTs typically specify a design rainfall event or a design storm, which is then converted to a synthetic design hydrograph by standard hydrologic techniques or by appli- cation of a rainfall-runoff model. The resulting event-based hydrograph is used to determine the BMP design parameters that will achieve the required percent capture, detention time, and peak flow attenuation. Hence, the BMP is designed to a single condition that may not have been observed and does not take into account BMP performance under a full range of hydrologic conditions. Continuous Simulation Analysis: Continuous hydrologic simulation has emerged over the last 30 years as a more robust alternative to event-based simulations for assessing the perfor- mance of BMPs (WEF, 1998; Strecker et al., 2005; Oregon State University et al., 2006). Results from continuous hydrologic simulations are based on the observed long-term precipita- tion patterns that are more representative of the variety of hydrologic conditions that affect runoff response and BMP performance. The continuous model generates a physi- cally based long-term runoff hydrograph by accounting for changes in soil moisture, infiltration, depression storage, and the long-term precipitation pattern. In some areas they also can include snow and snowmelt effects as well. Model- ing BMP performance in response to the long-term hydro- graph produces a more robust and comprehensive analysis of expected operational conditions than is possible from dis- crete, single-event models. Continuous simulation using the USEPA Storm Water Management Model (SWMM) was conducted to develop planning-level BMP sizing and design guidance for BMPs in an ultra-urban highway setting. The legacy SWMM4 was used for this effort instead of the current SWMM5 in order to utilize the plug-flow particle settling routine found in the legacy model that has yet to be incorporated into SWMM5. All other functionality is found in SWMM5 as well. SWMM was selected because of its ability to simulate (1) the long- term rainfall-runoff response of highway catchments, (2) the hydraulic performance of volume-based and flow- based BMPs, and (3) TSS settling in storage/detention units. S e c t i o n 6 BMP Sizing and Design

84 Figure 6.1 shows a conceptualization of the modeled ultra- urban highway catchment. The general modeling and eval- uation approach is as follows: 1. Define the catchment parameters. Catchment properties presented in Section 6.1.2 represent typical ultra-urban highway environments. 2. Compile appropriate precipitation data. Because pre- cipitation characteristics vary across the country and can affect BMP performance, sizing evaluation was conducted for 15 separate precipitation zones defined by Driscoll et al. (1989). Figure 6.2 shows the 15 rain zones and Section 6.1.3 describes the representative precipitation data. 3. Define representative BMPs for analysis. For this study the research team selected a generic non-proprietary detention vault to represent volume-based BMPs and a non-proprietary media filter system as the representative flow-based BMP. Sections 6.1.4 and 6.1.5 describe the BMP modeling approach used in the SWMM model, the pri- mary design parameters evaluated, and the performance criteria quantified. 4. Develop performance curves. The results from hydrologic simulations have been integrated into a spreadsheet sizing tool intended to provide planning-level sizing guidance that is included with this report. The tool enables users to explore tradeoffs between the primary sizing and design criteria and BMP performance. Section 6.1.6 describes the spreadsheet tool. 6.1.2 Highway Catchment Characteristics To capture the general characteristics of ultra-urban highway, a 1-acre paved highway catchment (100% imper- vious) as shown in Figure 6.1 was modeled. The catchment Figure 6.1. Continuous hydrologic simulation conceptualization. Source: After Driscoll et al. (1989) Figure 6.2. Hydrologic representation scheme for the United States.

85 width is 500 ft and the drainage length is 86 ft. This roughly corresponds to a six-lane highway section with 12 ft lanes, plus two 7 ft shoulders. Table 6.1 lists the catchment param- eters used in SWMM. 6.1.3 Precipitation Data To capture the effects of rapid precipitation responses that are typical of small highly impervious areas, a precipitation record with fine temporal resolution was required. Short- duration, high-intensity rainfall events typically control hydrologic engineering design for small, highly impervious urban catchments with short times of concentration (Tc). Therefore, precipitation recorded at a maximum of 5-minute intervals was desired. Continuous data sets for 1 and 5 min intervals are available online via the National Climatic Data Center (NCDC) for the period of 2000 through the present. The data are from the Automated Surface Observation System (ASOS) at locations within the United States. ASOS is a joint effort by the National Weather Service, Federal Aviation Administration, and the Department of Defense. The system has been collecting a full spectrum of surface climatic data at 1 and 5 min frequencies since the 1990s for a growing number of locations across the country. Hourly precipitation data tend to mask the sub-hour peak intensities. To demonstrate this, Figure 6.3 compares 5 min ASOS and hourly NCDC rainfall data for a single storm event in Boston, Massachusetts. In this example, if a flow-through BMP had been sized to be able to treat up to 0.2 in./h of runoff, it would have had a significant amount of bypass that would not have been accounted for by the use of hourly data. For ultra-urban highway drainage areas with characteristically short times of concentrations, finer resolution ASOS data are recommended for use in this BMP sizing modeling. Hence, the 5-min ASOS records were selected for this study, recognizing that a maximum duration of 10 years is placed on the continuous simula- tions, unless there are local data available from sources other than NCDC. Precipitation characteristics for each of the 15 rain zones are represented by available 5 min precipitation data from Pa ra me te r Va lu e Ar ea 1 ac re Im pe rv io us ne ss fr ac ti on 100% Wi dt h (h ig hw ay le ngth ) 500 ft Fl ow le ngth (h ig hw ay wi dt h) 87 ft (6 la ne s) Sl op e (w id th di re ct io n) 2% Im pe rv io us Ma nni ng ’s n 0. 013 Im pe rv io us depr e ssi on st or ag e 0. 02 in . Table 6.1. Highway catchment parameters used in SWMM. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 12 :0 0 12 :3 0 13 :0 0 13 :3 0 14 :0 0 14 :3 0 15 :0 0 15 :3 0 16 :0 0 16 :3 0 17 :0 0 17 :3 0 18 :0 0 18 :3 0 19 :0 0 19 :3 0 20 :0 0 20 :3 0 21 :0 0 21 :3 0 22 :0 0 22 :3 0 23 :0 0 23 :3 0 0: 00 Pr ec ip ita tio n (in ./h ) Time (h:min) ASOS 5-min and NCDC Hourly Rainfall Comparison for Boston Gage June 2, 2006 ASOS 5-min NCDC Hourly Figure 6.3. Precipitation intensities from NCDC hourly data and ASOS 5-min data.

86 a major metropolitan center selected within each zone. The ASOS data have a resolution of 0.01 in. Data spanning 2000 through 2009 (10 years) were used in the simulations. For this effort, snow accumulation and snowmelt processes were not modeled. Where snow accumulation and melt are impor- tant, it is recommended that they be incorporated. Table 6.2 presents the storm event statistics of selected stations repre- senting each of the 15 rainfall zones. The storm statistics are based on a 6 h interevent dry period that was used to separate precipitation data into storm events and exclude events less than or equal to 0.1 in. 6.1.4 Representation of Volume-Based BMPs BMP Scenario: Volume-based BMPs are represented by a generic non-proprietary rectangular extended-detention vault shown in Figure 6.4. This simple configuration can rep- resent a variety of BMPs with roughly similar storage and outlet configurations to the modeled facility, including for example surface extended-detention basins, underground vaults, and settling basins that are designed to promote sedi- mentation. Table 6.3 lists the geometry and conditions used to model the extended-detention basin with SWMM. When considering extended-detention facilities in ultra- urban applications, evaluation criteria include the volume and footprint of the facility and the sedimentation effi- ciency that can be achieved. These criteria are interdepen- dent. The settling of solids depends on the amount of time a given slug of water resides in the storage facility, as well as the size, shape, and specific gravity of the particle and the viscosity of runoff (temperature dependent). Basins with large storage capacity and longer detention times favor set- tling of particulates. However, a longer residence time increases the amount of bypassed or overflowed runoff due to exceed- ance of the basin capacity. Thus, the residence time needed to settle out particles represents a tradeoff between water Pr ec ip it at io n Zo ne Re pr es en ta ti ve Ci ty (A SO S Ga ge ID )* A nnu al No . of St or ms Av er ag e St or m Du ra ti on (h ) Av er ag e St or m Vo lu me (i n. ) Av er ag e St or m In te ns it y (i n. /h ) Av er ag e A nnu al Pr ec ip it at io n (i n. ) No rt he as t Bu ffa lo , NY (K BU F) 73 10. 6 0. 41 0. 055 29. 6 No rt he as t Co as ta l Bo st on , MA (K BO S) 64 11. 1 0. 56 0. 067 35. 8 Mi d- At la nt ic Wa sh in gton , DC (K DC A) 58 10. 1 0. 62 0. 094 36. 0 Ce nt ra l Na sh v ille , TN (K BN A) 66 8. 8 0. 63 0. 107 41. 3 No rt h Ce nt ra l Ch ic ag o, IL (K OR D) 56 9. 4 0. 60 0. 094 33. 5 So uthe as t At la nt a, GA (K AT L) 61 9. 0 0. 68 0. 109 41. 7 Ea st Gu lf Mi am i, FL (K MI A) 74 6. 3 0. 70 0. 166 52. 3 Ea st Te xa s Da lla s, TX (K DA L) 41 8. 8 0. 73 0. 125 29. 9 We st Te xa s Lu bbo ck , TX (K L BB) 29 7. 2 0. 54 0. 118 15. 6 So uthw es t Ph oeni x, AZ (K PH X) 15 7. 3 0. 52 0. 104 7. 8 We st In la nd La s Ve ga s, NV (K LA S) 11 7. 8 0. 55 0. 109 6. 2 Pa ci fi c So uthw es t Lo s An ge le s, CA (K LA X) 15 11. 3 0. 62 0. 075 9. 5 No rt hw es t In la nd Sa lt La ke Ci ty , UT (K SL C) 43 8. 4 0. 42 0. 062 18. 1 Pa ci fi c Ce nt ra l Sa n Fr an ci sc o, CA (K SF O) 31 13. 2 0. 56 0. 046 17. 2 Pa ci fi c No rt hw es t Se a ttl e, WA (K SE A) 78 12. 9 0. 62 0. 051 48. 3 * Du ra ti on of pr ec ip it at io n da ta an al yz ed : 2000–2009 Table 6.2. Storm event statistics of ASOS stations selected from 15 precipitation zones. Treated discharge L W Max depth = 4 feet Overflow outlet or spillway Untreated discharge Figure 6.4. Conceptual representation of volume-based BMPs.

87 quality and runoff capture. Continuous hydrologic simula- tions can specifically be used to evaluate this tradeoff. Design Parameters Evaluated: Storage volume and basin drain time are the design parameters evaluated in this study. • Storage volume: The modeled storage volume of the facil- ity varied over a wide range between 0.1 to 2.0 watershed in. For a fixed 4 ft depth, the corresponding footprints range from about 90 ft2 to 1800 ft2. The types of facilities that are represented range from small vault structures up to relatively large surface or underground detention systems. • Drain time: The full-depth drain time varied between 3 to 12 h, which is typical for space-constrained detention facilities. The outlet rating curve was specified to release the top half of storage in one-third of the drain time and the bottom half in two-thirds of the drain time (to maxi- mize small storm retention/treatment). Performance Criteria Quantified: BMP performance was quantified by the runoff capture and sediment capture as follows: • Average volume capture: The average capture volume is the fraction of total runoff that is detained and treated in the basin vs. the amount that overflows (i.e., the ratio of treated runoff to total runoff). Design specifications often require an 80% to 90% annual volume capture. Volume capture may be increased by enlarging the basin volume and/or by reducing the drawdown time. However, these strategies may correspondingly increase the required footprint and/ or reduce the ideal sedimentation efficiency. Note that for retrofit applications, it may be appropriate to design a sys- tem that captures less than typical requirements for new or redevelopment projects if space or other constraints exist or TMDLs or other goals can be met with a smaller facility. • Average sediment capture: Percent runoff capture is not a complete indicator of water quality performance in and of itself. High runoff capture is achieved with a larger out- let and rapid drawdown times at the expense of detention time that promotes sedimentation. Therefore, the research team quantified the sediment capture efficiency with the basic model of particle settling in SWMM4. Using Stokes Law, SWMM directly calculates the settling that takes place for each slug of water (plug flow) that is routed through the system. This analysis assumes that sedimentation processes in the BMP are represented by ideal settling theory and that sediment resuspension and washout are controlled through design features such as baffles and sediment traps. Thus, the estimated sediment capture represents the maximum sediment capture efficiency. Actual sediment removals will depend on site conditions and BMP design and would likely be lower than estimated efficiencies due to non-ideal set- tling and the potential effects of resuspension and washout. Note that sediment removal estimates may also be used to assess the removals of other pollutants that are associated with sediment. The following describes the procedures used to establish the modeled particle sizes and particle density. Total Suspended Solids Characteristics: Particle size distri- bution is an important design consideration for water quality treatment due to its influence on settling and because pollutant speciation is dependent on particle size (Wong et al., 2000). Literature information indicates there are large variations in PSDs in stormwater due to variations in site conditions, vehic- ular activity, wind patterns, rainfall/runoff characteristics, and the application of winter de-icing materials (Sansalone et al., 1998; Bent et al., 2003; Kim and Sansalone, 2008a). Thus, an assumed PSD used for the design of ultra-urban BMPs may often be significantly different from the actual PSD of sus- pended solids emanating from the site under investigation. Therefore, the distribution of particle size ranges (percentage by mass) is kept independent of the simulation runs. Particle settling analyses are conducted separately for a range of sedi- ment types and particle sizes, and the results for an arbitrary PSD are determined by integrating the individual results in a post-processing procedure. Five individual particle size ranges were modeled as shown in Table 6.4. The selected particle sizes are based on the AASHTO soil classification; however, the focus was on fine-grained sedi- ments that are difficult to remove in BMPs and are associated with higher concentrations of particulate pollutants. There- fore, silty-clayey grain sizes less that 75 µm were subdivided into four classifications, and a maximum grain size of 200 µm was assumed as larger sizes are easily settled and removed in BMPs. The particle specific gravity was based on average mea- surements of stormwater particulates (Li et al., 2008b; Krein and Schorer, 2000). Stormwater sediments exhibit a range of settling velocities and are usually significantly less dense than pure silica sands (specific gravity ~2.65). Table 6.4 also shows Pa ra me te r Va lu e Vo lu me 0. 1 to 2. 0 wa te rs hed in . or 363 to 7260 ft 3 (~ 2, 500 to 47, 000 ga llo ns ) Le ngth -t o- wi dt h ra ti o 4: 1 Ac ti ve st or ag e de pt h * 4 ft Su rf ac e ar ea De si gn vo lu me di vi ded by 4 ft (~ 90 to 1800 ft 2 ) Dr ai n ti me 3, 6, an d 12 h Ou tl et ra ti ng cu rv e To p ha lf in 1/ 3 of the dr ai n ti me Bo ttom ha lf in 2/ 3 of dr ai n ti me * De pt h ba se d on co mmon de si gn dept h fo r ur ba n BM Ps . A ddi ti on al de ad st or ag e ca n be in cl ud ed to he lp re du ce se di ment re su sp en si on . Table 6.3. Extended-detention basin parameters used in SWMM.

88 a representative PSD based roughly on the NJDEP BMP test- ing protocol. The effects of alternative PSDs may be estimated by users of the spreadsheet in a post-processing analysis that is included in the accompanying spreadsheet sizing tool. 6.1.5 Representation of Flow-Based BMPs BMP Scenario: A generic non-proprietary media filter sys- tem was chosen for the representative flow-based BMP because of the broad applicability of media filters to highway facilities. Media filters provide effective treatment for a range of highway pollutants, and they have flexible designs that are advantageous for surface and underground retrofit applications. Other types of flow-based BMPs, such proprietary cartridge filters and hydrodynamic separators, generally have well-defined sizing and design criteria specified by the manufacturer. Figure 6.5 shows the media filter system evaluated in this study. A 24-inch media bed with variable area was assumed, based on standard media bed designs. The long-term infiltra- tion capacity of the media bed controls the hydraulics of the flow-based BMP. The system includes active detention stor- age above the media, and untreated bypass occurs when the detention storage is exceeded. Discharge from the media bed collects in a freely draining underdrain system; however, the pore space in the media and underdrain were not included in the active storage. This simple system may represent sand fil- ters, organic filters, bioretention, and other media filters with similar media depth, area, detention storage, and conductiv- ity. Table 6.5 lists the geometry and conditions used to model the flow-based media filters. Design Parameters Evaluated: Evaluation criteria for media filtration in ultra-urban applications include the footprint of the facility and the runoff capture. The runoff capture depends on the hydraulic capacity of the facility, as well as the available detention storage. The hydraulic capacity depends largely on the media bed area and hydraulic conductivity, which are the fundamental design parameters evaluated in this study. The media composition was not considered in this study, as media selection is typically evaluated through column and treatability tests. Specific parameter values used in SWMM are as follows: • Media bed area: The media bed area was varied between 100 to 1000 ft2/acre, which corresponds to 0.23% to 2.3% of the tributary drainage area. • Detention storage: Detention storage is modeled with a fixed depth of 1 to 4 ft above the media bed (Urbonas, 2002). A pretreatment sedimentation wet vault, and the pore vol- ume within the media and underdrain are not included in the active storage. On a unit acre basis, the corresponding detention volume ranges from 100 to 4000 ft3. The effects of alternative ponding depths between 1 and 4 ft are estimated by interpolation in the spreadsheet sizing tool. • Flow-through rate: The flow-through rate is calculated from Darcy’s law as shown in Table 6.5. The maximum flow-through rate or infiltration capacity is constant when surface ponding is at the maximum height. For a fixed ponding depth and media thickness, the infiltra- tion capacity is controlled by the selection of the media hydraulic conductivity. • Media conductivity: The media conductivity was held constant in the simulations and was assumed to control the flow-through capacity of the media filter. Thus, the mod- eled design media conductivity represents a long-term infiltration capacity of the facility. The initial media con- ductivity is higher, but is expected to diminish over time due to clogging and surface crusting from fine particulates. Regular maintenance and periodic change-out of the media are required to retain the design infiltration capacity. The Se di me nt Ty pe Si ze ( µ m) Re pr es en ta ti ve % by Ma ss 1 Sp ec ific Gr av it y 2 Cl ay 1- 2 5 1. 2 to 1. 3 Fine silt 2- 8 15 1. 3 to 1. 35 Me di um s ilt 8- 32 20 1. 35 to 1. 4 Co ar se s ilt 32- 75 20 1. 4 to 1. 45 Fi ne sa nd 75- 200 40 1. 45 1 Re pr es enta ti ve si ze di st ri bu ti on a ppr ox im at el y ba se d on NJ DE P La bo ra to ry Te st Pr otoc ol (N JD EP , 2009a ) 2 Sp ec ific gr av it y ba se d on we t pa rt ic le sp ec if ic gr av it y me as ur ed by Li et al . ( 2008b ) Table 6.4. Particle size distributions and their properties used in simulations. Untreated Bypass Treated discharge Pre-treatment wet vault Media bed (2 feet) Active storage over media bed, 1 to 4 feet Underdrain (placed in 2-ft gravel base) Figure 6.5. Conceptual representation of flow-based BMPs.

89 maintenance frequency and time until clogging depends on the media area and sediment-loading rate. Note, an alterna- tive to using the media conductivity as the hydraulic control is to use a medium with high conductivity and design an outlet control orifice. This would help to reduce the onset of clogging. An outlet-controlled system with a known area can be approximately modeled within the spreadsheet tool by assuming a unit gradient (head-independent flow) and setting the hydraulic conductivity to the specific discharge Qoutlet/A, where Qoutlet is outlet-controlled discharge and A is the media bed area. Performance Criteria Quantified: BMP performance was quantified by the runoff capture and average media contact time as follows: • Average volume capture: The average capture volume was calculated from model results as the fraction of total runoff that percolates through the media bed. Darcy’s law estab- lishes the stage-discharge relationship used to calculate the discharge through the media bed in SWMM, taking into account the ponding head on the media bed as shown in Table 6.5. • Average media contact time: For filterable pollutants, lon- ger contact times generally correspond to greater pollutant removals. The contact time is controlled by the media area, depth, and porosity as shown in Table 6.5. In this analysis, the media depth and porosity is constant, such that con- tact time mainly depended on media area. Optimal contact times for the treatment of dissolved pollutants may range from less than 1 h to more than 10 h depending on the tar- get pollutant and media properties (Pitt and Clark, 2010). Longer contact time may not necessarily improve remov- als, as it could potentially enhance leaching of undesirable constituents from the media. 6.1.6 BMP Spreadsheet Sizing Tool The results of continuous simulation modeling studies have been synthesized into a Microsoft Excel®-based BMP sizing and design tool, which is included on a CD-ROM bound into this report. The purpose of the tool is to assist stormwater and highway professionals with planning-level sizing and design of detention and media filtration BMPs for ultra-urban highway runoff control. The tool generates BMP performance curves that relate the performance and design criteria described in the previous sections for each of the 15 rain zones. One of the significant features of the tool is that it allows users to explore BMP performance and retrofit siz- ing and design options based on selected design criteria and user-supplied inputs. 6.2 Sizing and Design of Detention-Based BMPs 6.2.1 Sizing Tool Example for Detention BMPs The Detention tab in the spreadsheet tool shows the sizing and design results for detention BMPs. There are four user- supplied inputs highlighted in yellow: • Drainage area: The drainage area is used to estimate the average annual runoff volume, detention volume, and TSS capture volume by proportional scaling of model results developed on a 1-acre catchment. Pa ra me te r Va lu e Me di a de pt h 2 ft Me di a ar ea 100 to 1000 ft 2 (0 . 23% to 2. 3% of the tr ib ut ar y wa te rs hed) De tent io n st or ag e vo lu me 1 to 4 ft ti me s th e medi a ar ea Lo ng -t er m medi a in f ilt ra ti on ca pa ci ty 1, 5, an d 50 in ./ h Fl ow -t hr ough ca pa ci ty 0. 002 to 1. 73 ft 3 /s De te rm in ed by Da rc y’ s la w: Q = KA J wh er e Q = di sc ha rg e K = hy dr au lic co nduc ti vi ty A = medi a be d ar ea J = hy dr au lic gr ad ie nt = (2 +h p) ÷2 hp = po nd in g he ig ht ab ov e the medi a be d Av er ag e medi a co nt ac t ti me De te rm in ed by 2 ft ÷ v av e wh er e v av e = av er ag e in te rs ti ti al ve lo ci ty = Q av e ÷ (A φ) Q av e = av er ag e f ilt er be d di sc ha rg e fr om SW MM A = filt er be d ar ea φ = medi a po ro si ty a ssu med to be 0. 35 Table 6.5. Media filtration characteristics and parameters used in SWMM.

90 • Average TSS concentration: Used to estimate the average annual sediment capture volume for varying detention volumes, drawdown rates, and assumed PSDs. • Particle size distribution: Defines the average mass frac- tion of clay, silt, and sand particles in runoff. The mass fraction must sum to 100%. The PSD is used to estimate the TSS capture as a function of basin size and drawdown time. • Rain zone: A dropdown menu listing the 15 rain zones. Representative storm statistics are listed for the selected rain zone. As an example, consider model results for a 1-acre catch- ment in the North Central rain zone, with an average TSS concentration of 100 mg/L. The North Central rain zone was modeled with precipitation data collected at the Chicago O’Hare Airport. Figure 6.6 shows the estimated runoff vol- ume captured as a function of detention volume and basin drain time. Results show that a basin volume in the range of 0.4 to 0.8 watershed inches (~1500 to 3000 ft3) will capture and treat about 85% of the annual runoff. An 85% capture approximately represents the knee of the curve (point of diminishing return) beyond which increasing the size of the basin achieves increasingly lower returns in additional runoff treated. Figure 6.6 also indicates that better volume capture occurs with faster drain times (DT) as expected in identically sized basins. Although volume capture improves with shorter detention time, this is not necessarily the case for sediment capture. Fig- ure 6.7 illustrates the change in TSS treatment performance based on the user input PSD. The left-hand plot shows the estimated sediment capture for a PSD dominated by sand and coarse silts. In this case, results suggest that 80% TSS capture can be achieved with a moderately sized basin and short drain times. In contrast, the right-hand plot shows estimated sedi- ment capture for a PSD dominated by finer silts and clays. For fine-grained sediment, better overall removals are obtained with longer detention times. However, an 80% capture is not achievable even at very large basin sizes. To achieve the 80% TSS performance standard with this PSD, a detention time much greater than 12 hours is required. Thus, alternative BMPs, such as wet vaults or media filtration, are likely more appropriate. 6.2.2 Evaluation of Batch Mode Operation The first flush phenomenon observed in highway runoff presents opportunities for stormwater treatment strategies (Stenstrom and Kayhanian, 2005). Fill-and-hold batch mode operation is a strategy that has been proposed and tested for improving the effectiveness of detention facilities. The higher residence time achieved by holding the runoff provides a greater opportunity for small particles to settle and reduces the potential for sediment resuspension and washout. Batch mode operation may be well suited for small underground vault systems, as it has been shown to effectively reduce sedi- 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 R un of f C ap tu re (% ) Detention Volume (watershed inches) Detention Facility - Runoff Capture North Central Rain Zone 3-hr DT 6-hr DT 12-hr DT Basin Drain Time Figure 6.6. Percent runoff captured for detention BMPs in the North Central rain zone. 3-hr DT 6-hr DT 12-hr DT Basin Drain Time Texture % mass Clay 5% Fine silt 5% Medium silt 10% Coarse silt 20% Sand 60% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Se di m en t C ap tu re (% ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Detention Volume (watershed inches) 3-hr DT 6-hr DT 12-hr DT Basin Drain Time Texture % mass Clay 10% Fine silt 20% Medium silt 40% Coarse silt 20% Sand 10% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Se di m en t C ap tu re (% ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Detention Volume (watershed inches) Figure 6.7. Effect of PSD on percent sediment capture for detention BMPs in the North Central rain zone.

91 ment resuspension and washout and improve sediment cap- ture (Landphair et al., 2007; Li et al., 2008a). Hydrologic simulation with SWMM was used to investi- gate and illustrate the potential improvement in overall sedi- ment capture with small vaults. SWMM was customized to simulate a dynamic controller that opens and closes an orifice outlet based on the following simple operating rules: • The basin outlet is closed when a lower-level depth sensor (e.g., 0.25 ft) is triggered. • The hold time is initiated when the upper-level depth sensor (full depth) is triggered. The outlet is automatically opened when the hold time is reached (1 to 3 h). To investigate the effectiveness of batch mode operation, event-based hydrograph and pollutograph data from the Caltrans first flush characterization study were used in the simulation study (Stenstrom and Kayhanian, 2005). The flow and pollutant monitoring data are shown in Figure 6.8 and clearly exhibit a first flush occurrence. These data were col- lected from a 3.2-acre ultra-urban highway catchment with nearly 100% impervious cover. Five different particle size ranges that represent the total TSS load in the runoff were simulated including 4–8 µm, 8–16 µm, 16–32 µm, 32–64 µm, and 64–128 µm. Based on this monitoring study, nearly 80% of the particles were observed to be less than 64 µm in size. Results of the batch mode modeling comparison are shown in Figure 6.9. Batch mode operation was evaluated for three small vault sizes—2000, 3000, and 4000 ft3—which correspond to about 0.17, 0.26, and 0.34 watershed inches. In each case, the outlet was sized for a 3 h drain time. Sediment removal was modeled with the ideal particle settling algorithms in SWMM, which does not account for sediment resuspension and wash- out. The TSS removal efficiencies shown in Figure 6.9 represent the percentage of sediment captured from the total sediment mass in the entire storm event. Source: Stenstrom and Kayhanian (2005) Figure 6.8. Flow and pollutant concentration from a first flush characterization study used in an event-based hydrologic simulation. 0% 10% 20% 30% 40% 50% 60% 70% 80% Non-Batch 1 h 2 h 3 h % S ed im en t R em ov al Holding Time (hours) 2000 CF 3000 CF 4000 CF Figure 6.9. Comparison of sediment removal performance between conventional and batch mode-operated detention vaults.

92 Simulation results shown in Figure 6.9 suggest that batch mode operation of small detention vaults can potentially improve sedimentation efficiency in comparison to conven- tional orifice outlets. Moreover, greater removal efficiency can be achieved with a smaller footprint facility operated in batch mode compared to a larger facility operated in con- ventional mode. For example, Figure 6.9 indicates that a batch mode-operated detention facility with a footprint of 2000 ft3 (~0.17 watershed inches) achieves 11% more sedi- ment removal efficiency than a conventionally operated facility with a volume of 4000 ft3 (~0.34 watershed inches). Model results also suggest there is an optimum holding time that depends on the size of the detention facility. For the smallest basin size (2000 ft3), the best removal efficiency is with the shortest holding time of 1 h because longer shut- in periods cause greater bypass losses. For the largest basin size (4000 ft3), a 2 h hold time provides the most efficient removal. For comparison, researchers at the Texas Trans- portation Institute found that a hold time of 3 h diminished the problems of resuspension and improved sediment cap- ture above 80% capture in controlled pilot tests (Landphair et al., 2007; Li et al., 2008a). Although batch mode operation is an emerging tech- nology, it is potentially an effective strategy for improving sedimentation efficiency in small underground vaults that are applicable in space-constrained settings. The spread- sheet sizing tool does not provide for a batch mode analy- sis. The user would need to examine this option through his/her own modeling effort. 6.3 Sizing and Design of Media Filtration-Based BMPs The Media Filter tab in the spreadsheet tool shows the sizing and design results for media filtration BMPs. There are five user-supplied inputs highlighted in yellow: • Drainage area: Determines the average annual runoff volume, filter area, and detention volume. • Average TSS concentration: Used in estimating the aver- age annual sediment-loading rate for varying filter areas. This is useful for estimating clogging rate and associated maintenance requirements. • Design media hydraulic conductivity: The long-term media conductivity will be achieved through ongoing maintenance and periodic media change-out. Results for the specified infiltration capacity are estimated by interpo- lation of the modeled results using infiltration rates of 1, 5, and 50 in./h. Thus, the specified infiltration must be in the range of 1 to 50 in./h. • Active detention storage depth: Active storage is assumed over the entire media filter area at depths between 1 and 4 ft. Results for intermediate depths are estimated by inter- polation of the modeled results at depths of 1 and 4 ft. • Rain zone: A dropdown menu listing the 15 rain zones. Rep- resentative storm statistics are listed for the selected rain zone. As an example, consider the results for a 1-acre catchment in the North Central rain zone, with an average TSS concen- tration of 100 mg/L, and a design infiltration capacity of 10 in./h. Results from the spreadsheet tool (shown in Fig- ure 6.10) illustrate the tradeoffs between runoff volume capture and filter area, media conductivity (K), and deten- tion depth. To achieve capture and treatment for 80% of the annual runoff, a filter area of about 2.0% of the tributary water- shed (~870 ft2/acre) is needed when the active storage depth is 1 ft. If sufficient head and storage are available, increasing the active storage to 4 ft will reduce the required filter area to about 0.75% of the tributary watershed (~325 ft2/acre). However, the TSS loading rate (mass per unit area) will more than double, increasing the clogging rate and maintenance requirements. Media filters can generally be expected to provide good effluent quality for particulate pollutants as supported by 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% R un of f C ap tu re (% ) Filter Area (% of tributary watershed) K = 1 in/hr K = 5 in/hr K = 50 in/hr K = 10 in/hr Design Infiltration Capacity 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% R un of f C ap tu re (% ) Filter Area (% of tributary watershed) Detention Storage = 1 ft Detention Storage = 4 ft K = 1 in/hr K = 5 in/hr K = 50 in/hr K = 10 in/hr Design Infiltration Capacity Figure 6.10. Percentage of runoff capture for media filtration in the North Central rain zone.

93 monitoring data from the BMP Database (Table 5.4). When media are tailored for removal of dissolved pollutants, the media contact time is an important design variable that is established through batch or column testing. Figure 6.11 shows the tradeoffs between average contact time and infil- tration capacity for a 2 ft thick media bed. Contact time is largely controlled by the infiltration rate, media thickness, and porosity. For this example, the average contact time ranges from less than 2 h to about 5 h at the design infiltra - tion capacity of 10 in./h. If shorter contact time is sufficient or is required to reduce leaching concerns, then the design media thickness can be reduced, which will also help to lower construction costs. The information in Figure 6.11 would need to be considered with that in Figure 6.10 (detention storage = 4 ft) to evaluate the appropriate sizing of the facil- ity. For example, a filter area at 1.5% of tributary area would treat 90% of the runoff with a K of 5 to 10 in./h, with an aver- age contact time of about 4 hours. This may be an adequate result, depending on the contact time needed. 6.4 Summary of Spreadsheet Sizing Tool The spreadsheet sizing tool is available on a CD-ROM bound into this report. Specific assumptions and user design variables are described in Sections 6.2.1 and 6.3 as well as in the tool itself. Modeling assumptions and caveats are given in those sections as well. While the spreadsheet comes to the user with protected cells, the user is free to unprotect all cells and view formulas. Similarly, the user is free to “unhide” the worksheets with SWMM results for each rainfall zone, from which the summary screening guidelines are derived by interpolation. The sizing tool is provided as a first-cut screening meth- odology for sizing guidance for BMPs in the ultra-urban highway setting. Users will naturally refine such guidance by using the spreadsheet output as good candidates for ini- tial BMP design using simulation models, which include the option of focusing on the user’s immediate location. Example SWMM4 input files are included for a representa- tive detention basin simulation and media filter simulation within the spreadsheet tool. The example SWMM files are included in a hidden worksheet, which can be viewed with the “Hide and Unhide” command in Excel. While not nec- essarily a simple task, it is at least a straightforward process to change SWMM input to desired local rainfall and evapo- transpiration data so that the user may focus the sizing tool to his/her specific location. 0 2 4 6 8 10 12 14 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% A ve ra ge C on ta ct T im e (h ou rs ) Filter Area (% of tributary watershed) Detention Storage = 4 ft K = 1 in/hr K = 5 in/hr K = 50 in/hr K = 10 in/hr Design Infiltration Capacity Figure 6.11. Average media contact time for media filtration in the North Central rain zone.