Long-Term Performance and Life-Cycle Costs of Stormwater Best Management Practices (2014)

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.

13 C H a P T E r 2 2.1 Background Since the NPDES program began in 1972 under Section 402 of the Clean Water Act, state DOTs have installed thousands of stormwater treatment facilities in the course of adding to the nation’s capacity to meet NPDES permit requirements and other similar state laws. These structural treatment facilities are commonly referred to as treatment BMPs and are designed to control the amount of stormwater runoff pollutants and volumes discharging to receiving waters. Maintenance of treatment BMPs is necessary to preserve their treatment and conveyance capacities as well as their intended water quality benefits. The EPA’s stormwater pollu- tion prevention fact sheet describes O&M plans as “an impor- tant part of a stormwater management program,” the goal of which is “to ensure that individual and interconnected storm- water BMPs continue to meet performance and design objec- tives” (U.S. EPA, 2010). 2.2 BMP Effectiveness and Long-Term Performance of BMPs 2.2.1 Most Commonly Used BMPs Treatment BMPs serve as permanent stormwater controls and typically include detention or retention ponds, con- structed wetlands, and sand filters (Urbonas, 1999; Carleton et al., 2000; Middleton and Barrett, 2008). Traditional storm- water practices are designed to retain runoff and release the water slowly after the storm event has passed. This helps to decrease the peak flow rates and flow volume and improves water quality through sedimentation and infiltration. The most common BMPs used by DOTs, based on the num- ber of states that use them, are basin configurations includ- ing wet ponds, detention ponds, and sediment or filtration basins (Eck et al., 2010). Other common approaches are veg- etated swales, filter strips, and infiltration systems. Table 2-1 indicates the percentages of state DOTs that use the listed BMPs in ultra-urban environments (Geosyntec Consultants et al., 2012). Vegetated slopes and roadside swales are more common BMPs in rural environments. Roadside ditches, if vegetated and designed with appropriate velocities for the water quality flow and of sufficient length, are BMPs and are described as “vegetated swales” in Table 2-1. Vegetated swales may be preceded by filter strips or vegetated areas that accept sheet flow. Some of the BMPs listed in Table 2-1, such as oil/ water separators, porous pavements, and cisterns, are used by DOTs at maintenance facilities but generally are not used for roadways. They are listed here since they were studied under a previous NCHRP project, but non-highway treatment BMPs are not included as a part of this project. 2.2.2 Summary of Previous Studies on BMP Performance This section summarizes previous studies on BMP perfor- mance with a primary focus on the International Storm- water BMP Database (BMPDB; http://www.bmpdatabase.org), which is the most comprehensive source of post-construction BMP performance data available in the world. The BMPDB is a collection of studies consisting primarily of BMP influent and effluent concentrations, rainfall event and runoff volumes, and ancillary test site and BMP information, such as water- shed characteristics and BMP design parameters. The BMPDB, as of January 2012, contained data from over 500 BMP stud- ies with over 15 structural BMP categories. Included within the BMPDB (in 2012) were 133 highway/roadway, park-and- ride, and maintenance station BMP research studies. These DOT studies were heavily focused on the west coast of the United States, with California accounting for approximately half of the studies (64 studies; 48%). Additionally, most of the research studies (77.5%) were highway and roadway related, followed by maintenance station (12%), then park-and-ride (10.5%). A summary of the research studies by state, BMP type, and land use can be found in Table 2-2. Because of the Literature Review and Survey Findings

14 breadth and depth of data contained in the BMPDB, it was a key resource used when developing the BMP spreadsheet tools. 2.2.2.1 BMP Constituent Removal The Water Environment Research Foundation (WERF) sponsored a comprehensive BMP technical report that included categorical performance assessments of all BMPs with suf- ficient water quality data in the BMPDB for statistical analy- sis (Geosyntec and Wright Water Engineers, 2012). The final version of that report was published in early 2012. It includes results for typical constituents of concern for state DOTs, such as total suspended solids (TSS), cadmium, copper, lead, nickel, zinc, phosphorus, and nitrogen. The median effluent concentration results are summarized in Tables 2-3 and 2-4. The tables indicate which BMPs and constituents had statis- tically significant differences between influent and effluent median concentrations. Total Suspended Solid Removal. All BMP types in the WERF study demonstrated statistically significant reductions in TSS concentrations and achieved median effluent concen- trations below 25 mg/L. Bioretention, media filters, and wet- land basins were shown to have the lowest median effluent TSS concentrations. Metals (Total and Dissolved). Most BMPs demonstrated significant reductions in total cadmium, copper, lead, and zinc concentrations, but the dissolved fractions of these metals are only significantly reduced by a handful of BMP types. While total metals include particles bound to sediment, and they can be removed through sedimentation and physical strain- ing, dissolved metals are mostly only removed through sorption and biochemical processes (Strecker et al., 2005). Therefore, BMPs expected to perform the best in dissolved metal concentration reductions provide adsorptive filtration or have long hydraulic residence times to allow for microbial transformations and plant uptake. NCHRP Report 767 (Barrett et al., 2013) explores methods for removing dissolved metals from urban runoff. Based on the available BMP data, vegetated strips show the best performance in removing dissolved metals (significant reduction in all dissolved metal effluent concentrations except for dissolved lead, which suffers from a high percentage of non-detects). Not enough studies (<3 studies) were available to evaluate the dissolved metal performance for bioretention, wetland basins, and wetland channels. Bioswales significantly reduced effluent concentrations for dissolved cadmium, dis- solved nickel, and dissolved zinc, but not for dissolved cop- per and dissolved lead. Swales are expected to provide similar performance to filter strips during small storms when flows are shallow and there is high contact with surface soils. How- ever, for larger storms, as the depth of flow increases, the con- tact area and contact time are reduced, thereby decreasing the removal efficiency, particularly for dissolved constituents. Nutrients (Phosphorus and Nitrogen). Retention ponds tend to perform the best in removing all forms of phosphorus BMP Surface detention (Dry ED/wet/infiltration basins, wetlands) 30 81 Vegetated/rock swales 29 78 Hydrodynamic separators 23 62 Oil/water separators 22 59 Infiltration trenches 18 49 Underground detention 17 46 Catch basin inserts 16 43 Low-impact development BMPs (e.g., bioretention, amended soils) 16 43 Proprietary media filters (e.g., storm filter) 15 41 Sand filters 14 38 Filter strips 14 38 Diversion to treatment facilities 10 27 Multichambered treatment train systems 7 19 Porous pavements 7 19 Cisterns 3 8 Notes: 37 DOTs responded to this survey. ED = extended definition. Source: Geosyntec Consultants et al., 2012 Number of States That Reported Using the BMP Percent of States That Reported Using the BMP Table 2-1. Percent of states reporting use of BMP type.

15 State and BMP Type BMP Study Count by Primary Land Use Total California 15 6 43 64 Biofilter – vegetated strip 2 – 29 31 Biofilter – vegetated swale 1 – 5 6 Detention basin (dry) – concrete or lined basin with open surface – – 1 1 Detention basin (dry) – surface vegetated-lined basin, empties after storm – – – – – 4 4 Filter – other media 1 1 Filter – peat/gravel mixed with sand 1 2 3 Filter – sand 3 3 1 7 Manufactured device 7 1 2 10 Retention pond (wet) - surface pond with a permanent pool – – 1 1 Delaware – – 9 9 Bioretention – – 1 1 Filter – sand – – 1 1 Manufactured device – – – – 7 7 Florida 7 7 Biofilter – vegetated swale – – 6 6 Retention pond (wet) – surface pond with a permanent pool – – 1 1 North Carolina – – 14 14 Biofilter – vegetated strip – – 2 2 Biofilter – vegetated swale – – 2 2 Biofilter – wetland vegetation swale – – 2 2 Composite – overall site BMP – – 4 4 Permeable friction course – – 3 3 Porous pavement – porous asphalt – – 1 1 Oregon – – 1 1 Manufactured device – – 1 1 Pennsylvania 1 – – 1 Manufactured device 1 – – 1 Texas – – 16 16 Biofilter – vegetated strip – – 2 2 Biofilter – vegetated swale – – 1 1 Detention basin (dry) - concrete or lined basin with open surface – – 1 1 Composite – overall site BMP – – 1 1 Control – no BMP/control site – – 2 2 Filter – gravel – 1 1 Filter – sand – – 3 3 Manufactured device – – 1 1 Permeable friction course – – 3 3 Retention pond (wet) – surface pond with a permanent pool – – 1 1 Virginia – 6 12 18 Biofilter – vegetated swale – – 9 9 Biofilter – vegetated strip – – 1 1 Detention basin (dry) – surface vegetated-lined basin, empties after storm – 4 – 4 Wetland – basin with open water surfaces – – 1 1 Wetland – basin without open water (wetland meadow) – – 1 1 Composite – overall site BMP – 2 2 Washington – – 1 1 Bioretention – – 1 1 Grand Total 16 12 103 131 Maint. Station Park and Ride Roads/ Highway Table 2-2. Summary of transportation-related BMPDB research studies (2012).

16 BMP Type TSS, mg/L (95% CI)a Dissolved Cadmium µg/L (95% CI)a Total Cadmium µg/L (95% CI)a Dissolved Copper µg/L (95% CI)a Total Copper µg/L (95% CI)a Dissolved Lead µg/L (95% CI)a Total Lead µg/L (95% CI)a Dissolved Nickel µg/L (95% CI)a Total Nickel µg/L (95% CI)a Vegetated strip 19.1 (16.0, 21.5) 0.09 (0.07, 0.11) 0.18 (0.09, 0.20) 5.40 (4.50, 5.90) 7.30 (6.40, 7.90) 0.26 (0.19, 0.35) 1.96 (1.30, 2.20) 2.09 (2.00, 2.15) 2.92 (2.40, 3.10) Bioretention 8.3 (5.0, 9.0) N/Ad 0.94 (0.25, 1.00) N/Ad 7.67 (4.60, 9.85) N/Ad 2.53 (2.50, 2.50) N/Ad N/Ad Bioswale 13.6 (11.8, 15.3) 0.12 (0.09, 0.15) 0.31 (0.27, 0.34) 8.02 (6.30, 9.24) 6.54 (5.70, 7.70) 1.08 (0.76, 1.60) 2.02 (1.80, 2.29) 2.04 (2, 2.40) 3.16 (2.30, 4.20) Composite 17.4 (12.4, 18.8) N/Ad 0.50 (0.43, 0.50) 5.00 (5.00, 5.00) 5.88 (5.05, 6.79) 0.29 (0.09, 0.44) 4.78 (3.00, 5.61) N/Ad N/Ad Detention basin 24.2 (19.0, 26.0) 0.50b (0.50, 0.50) 0.31 (0.25, 0.35) 3.52 (2.80, 4.72) 5.67 (4.00, 6.80) 0.66 (0.48, 0.90) 3.10 (2.15, 4.30) 2.55 (2.00, 3.00) 3.35 (2.20, 3.75) Manufactured devicee 18.4 (15.0, 19.9) 0.30 (0.24, 0.39) 0.28 (0.20, 0.31) 6.08 (4.82, 7) 10.16 (7.94, 11.0) 1.24 (1.00, 1.38) 4.63 (3.80, 5.16) 1.92 (0.44, 2.00) 4.51 (3.11, 5.00) Media filter 8.7 (7.4, 10.0) 0.18 (0.11, 0.20) 0.16 (0.10, 0.20) 4.35 (3.58, 5.10) 6.01 (5.10, 6.60) 1.00 (1.00, 1.00) 1.69 (1.30, 2.00) 1.90 (0.99, 2.00) 2.20 (2.00, 2.60) Porous pavement 13.2 (11.0, 14.4) 0.04c (0.02, 0.05) 0.25c (0.25, 0.25) 5.75 (4.90, 5.91) 7.83 (6.80, 8.10) 0.50c (0.50, 0.50) 1.86 (1.38, 2.21) 0.43c (0.33, 0.52) 1.71 (1.40, 1.80) Retention pond 13.5 (12.0, 15.0) 0.10 (0.07, 0.13) 0.23 (0.20, 0.29) 4.24 (4.00, 4.57) 4.99 (4.06, 5.00) 0.48 (0.23, 0.96) 2.76 (2.00, 3.00) 2.11 (1.40, 2.53) 2.19 (2.00, 2.60) Wetland basin 9.06 (7.0, 10.9) N/Ad 0.18 (0.10, 0.20) N/Ad 3.57 (3.00, 4.00) N/Ad 1.21 (1.00, 1.55) N/Ad N/Ad Wetland channel 14.3 (10.0, 16.0) N/Ad 0.49 (0.19, 0.50) N/Ad 4.81 (3.61, 5.20) 0.52 (0.12, 0.75) 2.49 (1.40, 3.11) N/Ad 2.18 (2.00, 2.40) Notes: (Bolded and italicized to show statistically significant decrease between influent and effluent median concentrations.) CI = confidence interval. a. Computed using the bias corrected and accelerated (BCa) bootstrap method by Efron and Tibishirani (1993). b. Hypothesis testing shows statistically significant increases for this BMP category. c. Conclusions are limited for this BMP category due to a large percentage of non-detects in the influent. d. N/A – not available or fewer than three studies for BMP/constituent. e. “Manufactured device” includes cartridge filters, inlet inserts, oil/grit separators, and hydrodynamic separators. Source: Geosyntec Consultants and Wright Water Engineers, 2012. Table 2-3. BMP median effluent concentration for constituents commonly reported in the BMPDB (continued as Table 2-4). and nitrogen, followed by wetland basins. These practices include a permanent pool, which increases the hydraulic residence time, allowing sedimentation and biochemical processes to take place while also having both aerobic and anaerobic zones to facilitate oxidation-reduction processes (e.g., nitrification and denitrification). In general, the veg- etated strip, bioretention, bioswale, and wetland channel do not show a statistically significant decrease in concentrations, and some sites can show increases in phosphorus concentra- tions. Leaching of phosphorus from soils and planting media and resuspension or degradation of captured particulate phosphorus may be a cause of the increases observed. If soil amendments contain high concentrations of phosphorus (e.g., compost), the phosphorus could be released into the BMP effluent. 2.2.2.2 DOT BMP Constituent Long-Term Removal Studies Unfortunately, the BMPDB contains few studies with long- term data sets—most studies span only 1 to 2 years. Studies containing more than a 4-year monitoring record constitute only 28 (21%) of the 133 DOT studies within the BMPDB. Among these 28 studies, 20 pertained to vegetated strips, three pertained to sand filters, three pertained to porous asphalt, one pertained to bioretention, and one pertained to vegetated swales. Sites featuring some of the most commonly used transportation-related BMPs, such as swales, vegetated strips, and sand filters, are of special interest for longer-term study. Additionally, sites with new and innovative transporta- tion BMPs, including PFC pavements and the Washington State DOT (WSDOT) ecology embankment, are also of spe- cial interest. For this reason, 10 of the 28 DOT studies were selected for further review based on the type and location of the BMP. At least one of each BMP type available in the DOT studies was chosen from highway studies containing more than 4 years of data. They were evaluated here to determine if performance changed over time and, if so, whether BMP life should be considered in the development of the BMP Evaluation Tool. Based on the analysis of the limited data sets (described in the following), it was determined that there was

17 BMP Type Dissolved Zinc µg/L (95% CI)a Total Zinc µg/L (95% CI)a Total Phosphorus mg/L (95% CI)a Orthophosphate mg/L (95% CI)a Dissolved Phosphorus mg/L (95% CI)a Total Nitrogen mg/L (95% CI)a Total Kjeldahl Nitrogen mg/L (95% CI)a NOx as Nitrogen mg/L (95% CI)a Vegetated strip 14.0 (10.0, 16.0) 24.3 (16.0, 26.0) 0.18b (0.15, 0.20) 0.06b (0.04, 0.07) 0.25b (0.16, 0.26) 1.13 (1.00, 1.23) 1.09 (0.97, 1.12) 0.27 (0.24, 0.31) Bioretention N/Ad 18.3 (7.7, 25.0) 0.09 (0.07, 0.10) 0.04b (0.02, 0.05) 0.13 (0.05, 0.18) 0.90 (0.74, 0.99) 0.60 (0.46, 0.72) 0.22 (0.19, 0.25) Bioswale 24.5 (21.3, 27.5) 22.9 (20.0, 26.6) 0.19b (0.17, 0.20) 0.12b (0.10, 0.13) 0.07b (0.05, 0.11) 0.71 (0.63, 0.82) 0.62 (0.50, 0.70) 0.25 (0.20, 0.28) Composite 9.9 (4.4, 10.0) 33.0 (28.5, 39.5) 0.13 (0.11, 0.15) 0.07 (0.04, 0.10) 0.08 (0.06, 0.09) 1.71 (1.45, 1.81) 102 (0.88, 1.14) 0.40 (0.33, 0.46) Detention basin 11.08 (8, 17) 29.7 (17.1, 38.2) 0.22 (0.19, 0.24) 0.39 (0.24, 0.56) 0.11 (0.08, 0.12) 2.37b (1.75, 2.69) 1.61 (1.16, 1.78) 0.36 (0.24, 0.45) Manufactured devicee 53.3 (44.0, 64.0) 58.5 (52.8, 63.5) 0.12 (0.10, 0.13) 0.10 (0.06, 0.13) 0.06 (0.04, 0.07) 2.22 (1.90, 2.41) 1.48 (1.32, 1.55) 0.41 (0.35, 0.44) Media filter 12.2 (8.3, 17.0) 17.9 (15.0, 20.0) 0.09 (0.08, 0.10) 0.03 (0.02, 0.03) 0.08 (0.06, 0.09) 0.82 (0.68, 0.99) 0.57 (0.50, 0.61) 0.51b (0.46, 0.57) Porous pavement 6.5 (4.9, 7.9) 15.0 (12.5, 16.8) 0.09 (0.08, 0.09) 0.05 (0.04, 0.06) 0.05 (0.04, 0.05) 1.49 (1.28, 1.65) 0.80 (0.74, 0.90) 0.71b (0.59, 0.77) Retention pond 9.6 (5.3, 10.9) 21.2 (20.0, 23.0) 0.13 (0.12, 0.14) 0.04 (0.03, 0.05) 0.06 (0.06, 0.07) 1.28 (1.19, 1.36) 1.05 (0.98, 1.10) 0.18 (0.15, 0.20) Wetland basin N/Ad 22.0 (16.7, 24.3) 0.08 (0.07, 0.09) 0.02 (0.01, 0.02) 0.05 (0.03, 0.06) 1.19 (1.04, 1.21) 1.01 (0.92, 1.09) 0.08 (0.05, 0.11) Wetland channel 9.5 (2.9, 10.0) 15.6 (11.0, 20.0) 0.14 (0.13, 0.17) 0.06b (0.04, 0.06) 0.09 (0.07, 0.10) 1.33 (1.05, 1.56) 1.23 (1.10, 1.30) 0.19 (0.15, 0.22) Notes: (Bolded and italicized to show statistically significant decrease between influent and effluent median concentrations.) CI = confidence interval. a. Computed using the bias corrected and accelerated (BCa) bootstrap method by Efron and Tibishirani (1993). b. Hypothesis testing shows statistically significant increases for this BMP category. c. Conclusions are limited for this BMP category due to a large percentage of non-detects in the influent. d. N/A – not available or fewer than three studies for BMP/constituent. e. “Manufactured device” includes cartridge filters, inlet inserts, oil/grit separators, and hydrodynamic separators. Source: Geosyntec Consultants and Wright Water Engineers, 2012. Table 2-4. BMP median effluent concentration for constituents commonly reported in the BMPD (continued). no basis for adjusting performance based on BMP life. The BMPs evaluated were: • Ecology embankment in Washington (one study), • Vegetated swale in Texas (one study), • Vegetated strip in California (four studies), • Sand filter in California (three studies), and • Permeable friction course overlay in Texas (one study). Ecology Embankment Study. The ecology embankment study was conducted in Auburn, WA [BMP name: WA Ecology Embankment at SR (State Route) 167, MP (milepost) 16.4]. The ecology embankment is a special type of bioretention design where sheet flow runoff from the adjacent pavement surface is filtered via interflow along an engineered slope and then collected in an underdrain at the toe of the slope. Essen- tially, the ecology embankment is a hybrid between a filter strip and a bioretention cell containing a custom filtration media mix. The filtration media mix consists of crushed rock (screened between ³⁄8-in. and #10 sieve) and three amend- ments: dolomite, gypsum, and perlite (WSDOT, 2006). Pre- treatment consists of a vegetated strip between the paved shoulder and the filtration media. WSDOT sponsored this study to analyze seven constituents of concern from August 21, 2001, to April 7, 2005. The ecology embankment was located on the shoulder of northbound SR 167 treating runoff from an approximate 0.5-acre drainage area. The drainage area con- sisted of two lanes of traffic and two shoulders. Table 2-5 compares the influent and effluent medians for the ecology embankment to the 2012 categorical performance estimates for bioretention and vegetated strips for the entire BMPDB. As shown in the table, the median effluent concen- trations for the ecology embankment were lower than the cate- gorical performance estimates for total suspended solids and total phosphorus despite having higher median influent con- centrations. The low total phosphorus effluent concentration achieved is likely due to sorption and precipitation of phos- phorus promoted by the dolomite and gypsum amendments

18 and the lack of any organic material in the ecology mix. While the ecology embankment effluent concentration was higher for total copper, the influent concentrations were also gener- ally much higher (a median influent of 52 µg/L compared to 17 µg/L for the bioretention category and 25 µg/L for the veg- etated strip category). The dissolved copper performance for the ecology embankment is not significantly different from that for vegetated strips. Vegetated Swale Study, Texas. The vegetated swale study was conducted in Austin, Texas (BMP name: Brodie Lane Swale) by the City of Austin. Vegetated swales may accept flow along the entire length of the swale or may only accept flow at the upstream end, as was the case for this study. Perfor- mance is theoretically better for swales that do not operate with spatially varied flow, but in practice, the many variables that affect swale performance (depth, velocity, soil perme- ability) may overshadow the flow condition. The Texas DOT sponsored a highway vegetated swale study analyzing 16 constituents of concern from May 1, 2000, to May 1, 2005. The vegetated swale received runoff from an approximately 0.5-acre drainage area. The drainage area consisted of the eastern portion of Brodie Lane in Austin, Texas. (Other spe- cific design information is missing from the BMPDB.) Two distinct monitoring periods occurred at this site: one over the 2000–2001 wet season and one over the 2004–2005 wet season. Table 2-6 provides a comparison of median influ- ent and effluent concentrations for the Brodie Lane swale to the categorical BMP performance for TSS, dissolved copper, total copper, total Kjeldahl nitrogen (TKN), orthophosphate, and total phosphorus. The median effluent concentrations for the Brodie Lane swale were lower than the categorical performance estimates for total copper, TKN, and total phosphorus. However, the median TSS effluent concentra- tion for the Brodie Lane swale was more than twice the cate- gorical median performance. While the Brodie Lane swale had higher median influent TSS concentrations, the BMP was unable to achieve statistically significant reductions in TSS. Consistent with the categorical BMP performance, the Brodie Lane swale did not achieve statistically significant removal of TKN and tended to increase total phosphorus concentrations. TSS (mg/L) Dissolved Copper (µg/L) Total Copper (µg/L) TKN (mg/L) Dissolved Phosphorus (mg/L) Total Phosphorus (mg/L) WA ecology embankment at SR 167 MP 16.4 In Out In Out In Out In Out In Out In Out 96 5 12.5 7.1 52 10 N/A N/A N/A N/A 0.21 0.04 Categorical BMP Performance (Geosyntec Consultants and Wright Water Engineers, 2012) Bioretention 37.5 8.3 N/A N/A 17 7.67 0.94 0.6 0.25 0.13 0.11 0.09 Vegetated strip 43.1 19.1 11.66 5.4 24.52 7.3 1.29 1.09 0.08 0.25 0.14 0.18 Notes: Bolded and italicized values indicate effluent median that is statistically significantly less than the influent median (alpha = 0.05). TKN = total Kjeldahl nitrogen. Table 2-5. Median concentrations of WA ecology embankment bioretention compared to Geosyntec Consultants and Wright Water Engineers (2012) categorical BMP performance summaries. Table 2-6. Median concentrations from flow-weighted composite samples of Brodie Lane swale compared to Geosyntec Consultants and Wright Water Engineers (2012) categorical BMP performance summaries. TSS (mg/L) Dissolved Copper (µg/L) Total Copper (µg/L) TKN (mg/L) Orthophosphate as Phosphorus (mg/L) Total Phosphorus (mg/L) In Out In Out In Out In Out In Out In Out Brodie Lane swale 56.0 43.3 N/A N/A 3.65 3.0 0.6 0.5 N/A N/A 0.11 0.12 Categorical BMP Performance (Geosyntec Consultants and Wright Water Engineers, 2012) Bioswale 21.7 13.6 11.01 8.02 10.9 6.54 0.72 0.62 0.06 0.07 0.11 0.19 Notes: Bolded and italicized values indicate effluent median is statistically significantly less than the influent median (alpha = 0.05). Italicized-only values indicate effluent median is statistically significantly greater than the influent median (alpha = 0.05).

19 Vegetated Strip Studies, California. Vegetated strip studies in California included: 1. Moreno Valley, CA (BMP name: Moreno Valley 2, aver- age annual rainfall: 9.9 in.). Caltrans sponsored a highway vegetated strip study analyzing 27 constituents of concern from November 24, 2001, to May 22, 2006. The vegetated strip treated runoff from an approximately 0.1-acre drain- age area. The drainage area consisted of an asphalt-paved eight-lane highway (eastbound Moreno Valley freeway). Three other vegetated strip studies were conducted at this location (Moreno Valley 3, Moreno Valley 4, and Moreno Valley 5). 2. Redding, CA [BMP name: Redding RVTS (roadside veg- etated treatment study) 2.2 m, average annual rainfall: 34.6 in.]. Caltrans sponsored a highway vegetated strip study analyzing 31 constituents of concern from Novem- ber 11, 2001, to February 23, 2008. The vegetated strip treated runoff from an approximately 0.07-acre drainage area. The drainage area consisted of an asphalt four-lane highway (eastbound 299 between Chum Creek and Old Oregon Trail). Two other vegetated strip studies were con- ducted at this location (Redding RVTS 4.2 m and Redding RVTS 6.2 m). 3. Sacramento, CA (BMP name: Sacramento RVTS 2, aver- age annual rainfall: 21.1 in.). Caltrans sponsored a high- way vegetated strip study analyzing 31 constituents of concern from November, 12, 2001, to February 19, 2008. The vegetated strip treated runoff from an approximately 0.1-acre drainage area. The drainage area consisted of an asphalt-paved six-lane highway (northbound of I-5 north of the Laguna St. exit). Three other vegetated strip studies were conducted at this location (Sacramento 3, Sacramento 4, and Sacramento 5). 4. Yorba Linda, CA (BMP name: Yorba Linda RVTS 2, aver- age annual rainfall: 14.4 in.). Caltrans sponsored a high- way vegetated strip study analyzing 29 constituents of concern from November 24, 2001, to February 21, 2008. The vegetated strip treated runoff from an approximately 0.2-acre drainage area. The drainage area consisted of an asphalt-paved 13-lane highway (Riverside Freeway at Woodcreek). Three other vegetated strip studies were con- ducted at this location (Yorba Linda RVTS 3, Yorba Linda RVTS 4, and Yorba Linda RVTS 5). Each of these sites is located in a dry summer subtropi- cal or Mediterranean climate. As shown in Table 2-7, Sacra- mento, Moreno Valley, and Yorba Linda vegetated strips did not achieve the categorical bioswale median effluent concen- trations for any of the constituents analyzed, but the Redding vegetated strip did meet the effluent concentrations for TSS, dissolved copper, total copper, and TKN. All studies increased orthophosphate and total phosphorus, which is similar to what has been observed for the overall BMPDB. The sam- pling locations were at 1.1, 2.6, 2.3, and 2.2 m from the edge of the pavement for the Sacramento, Moreno Valley, Yorba Linda, and Redding studies, respectively. Additionally, Sac- ramento was located on hydrologic soil type D, and Moreno Valley and Yorba Linda vegetated strips were located on steep slopes (>10%), both factors influencing and limiting infiltra- tion within the vegetated strip. Inhibiting infiltration reduces concentration reductions associated with sedimentation and particle retention due to greater flow depth and velocity in the strip. Table 2-7. Median concentrations of Moreno Valley 2, Redding RVTS 2.2 m, Sacramento RVTS 2, and Yorba Linda RVTS 2 compared to Geosyntec Consultants and Wright Water Engineers (2012) categorical BMP performance summaries. TSS (mg/L) Dissolved Copper (µg/L) Total Copper (µg/L) TKN (mg/L) Orthophosphate as P (mg/L) Total Phosphorus (mg/L) In Out In Out In Out In Out In Out In Out Moreno Valley 2 61.5 83 20 17.5 41.5 28.5 2.15 1.8 0.08 0.14 0.29 0.35 Redding RVTS 2.2 m 21.5 8 2.2 2.3 3.95 3.5 0.87 0.58 0.01 0.02 0.04 0.06 Sacramento RVTS 2 50 27 5.6 5.75 15 12 1.2 1.25 0.12 0.21 0.26 0.31 Yorba Linda RVTS 2 64 100 15 14 37 44 1.6 2.1 0.06 0.06 0.22 0.32 Categorical BMP Performance (Geosyntec Consultants and Wright Water Engineers, 2012) Vegetated strip 43.1 19.1 11.7 5.4 24.5 7.3 1.29 1.09 0.08 0.25 0.14 0.18 Notes: Bolded and italicized values indicate effluent median is statistically significantly less than the influent median (alpha = 0.05). Italicized-only values indicate effluent median is statistically significantly greater than the influent median (alpha = 0.05).

20 Moreno Valley 2 also contained the shortest vegetation, approximately 3 cm high, and the lowest vegetation coverage— average vegetation cover of less than 15%. Dense vegetation decreases runoff velocities and increases the opportunity for straining of particles, facilitating sedimentation and reduc- tion of constituents. The Redding vegetated strip had at most 85% vegetation cover throughout the study, and the height of the vegetation was at least 5 cm, with the tallest vegetation of 28 cm. The shortest effective lengths of vegetated strips were 4.6 m, 13 m, 0 m (edge of pavement), and 4.2 m for Sacra- mento, Yorba Linda, Moreno Valley, and Redding, respectively (Caltrans, 2003d). Redding performed the best in terms of median effluent concentrations and had the smallest effective length. Sacramento was the only strip that achieved statisti- cally significant removal of total copper. During the Caltrans (2003d) study, this site had high vegetative cover (80%–98%), which likely influenced the retention effectiveness at this site. Sand Filter Studies, California. Sand filter studies in California included: 1. Redding, CA (BMP name: Mountain Gate Partial Sedimen- tation Austin Sand Filter). The California DOT (Caltrans, 2003a) sponsored a highway sand filter study analyzing 33 constituents of concern from February 6, 2002, to February 6, 2006. The sand filter treated runoff from an approximately 2.5-acre drainage area. The drainage area consisted of a four-lane highway (northbound and south- bound I-5 near the Mountain Gate exit). 2. Shasta, CA (BMP name: Mt. Shasta Maintenance Station Sand Filter). Caltrans sponsored a maintenance station study analyzing 32 constituents of concern from Novem- ber 7, 2002, to April 15, 2006. The sand filter treated runoff from an approximately 2.6-acre drainage area. The drainage area consisted of a DOT maintenance station in Shasta, CA. 3. Whittier, CA [BMP name: Eastern SF (sand filter)]. Caltrans sponsored a maintenance station study analyzing 27 con- stituents of concern from November 11, 2001, to April 19, 2007. The sand filter treated runoff from an approximately 1.5-acre drainage area. The drainage area consisted of a DOT maintenance station in Whittier, CA. The California sand filter BMPs contain only sand, result- ing in higher filtration rates as compared to media filters with blended compost media. Sand-only filters provide good removal of suspended solids and any constituents bound to particles, but typical filtration sand is relatively inert, and it would not be expected to reduce dissolved constituents unless an organic biofilm develops within the media bed. Biofilms may only develop in wet climates (all of these sites were rela- tively dry climates) where the media bed does not completely dry out between storms. Data from each of the sand filters are compared to the categorical performance estimates for media filters for the entire BMPDB. The Redding sand filter is a partial sedimentation sand filter that does not include a sedimentation forebay, using one basin for both sedimentation and sand filtration. The Mt. Shasta maintenance station sand filter and the Whittier sand filter are full-sedimentation sand filters with dedicated basins for sedimentation that are separate from the sand filtration basin. Table 2-8 compares the influent and effluent median val- ues for the three sand filters to the categorical performance estimates for media filters for the entire BMPDB. As shown in the table, both the influent and effluent concentrations for the Mountain Gate sand filter were lower than the cate- gorical performance estimates for all the constituents in the table. The comparatively lower constituent influent concen- trations at the Mountain Gate sand filter may be responsible for some of the differences in effluent quality between the Mountain Gate sand filter and the categorical performance TSS (mg/L) Dissolved Copper (µg/L) Total Copper (µg/L) TKN (mg/L) Orthophosphate as P (mg/L) Total Phosphorus (mg/L) Mountain Gate SF In Out In Out In Out In Out In Out In Out 35.2 4.3 3.6 1.35 8.8 2.65 0.78 0.3 0.03 0.02 0.09 0.05 Mt. Shasta SF 18 1 0.89 1.1 4 1.4 0.3 0.16 0.02 0.01 0.05 0.01 Eastern SF 44 11 5.4 6.45 13 7.5 0.87 0.57 N/A N/A 0.13 0.09 Categorical BMP Performance (Geosyntec Consultants and Wright Water Engineers, 2012) Media filter 52.7 8.7 5.37 4.35 11.28 6.01 0.96 0.57 0.05 0.03 0.18 0.09 Notes: Bold and italicized values indicate effluent median is statistically significantly less than the influent median (alpha = 0.05). Italicized values indicate effluent median is statistically significantly greater than the influent median (alpha = 0.05). Table 2-8. Median concentrations of Mountain Gate sand filter compared to Geosyntec Consultants and Wright Water Engineers (2012) categorical BMP performance summaries.

21 estimates. Other factors may be maintenance and climatic differences. As shown in Table 2-8, both the median influent and efflu- ent concentrations for the Mt. Shasta station sand filter were lower than the categorical performance estimates for all the constituents in the table. However, with the exception of TSS, the sand filter did not show a statistically significant differ- ence between the effluent and the influent for any of the con- stituents. This site had relatively low influent concentrations compared to other sand filters in the BMPDB, with many of the constituents at or near the analytical detection limits. Unlike the Mountain Gate and Mt. Shasta sand filters, the Eastern SF shows higher effluent concentrations for all con- stituents except TKN and total phosphorus compared to the categorical standards. The relatively poor performance of the Eastern SF compared to the Shasta and Mountain Gate sand filters is likely related to loading. The influent concen- trations are significantly higher for the Eastern SF, resulting in higher effluent concentrations. Other factors, such as rainfall intensities, pretreatment designs, and site characteristics, could be explored for each study to better understand the differences in median effluent concentrations between all the sites explored and the media filters in the overall BMPDB. Permeable Friction Course Study, Texas. The permeable friction course study was conducted in Austin, Texas (BMP name: AustinTX1PFC). PFC was installed on the southbound loop of 360, approximately 1.5 km north of Lakewood Drive. The study analyzed 11 constituents of concern from April 1, 2004, to September 1, 2009, and received runoff from south- bound 360. PFC is an innovative roadway material placed in an approx- imately 25 to 50 mm overlay on top of regular pavement. PFC improves safety and driving conditions by allowing the road surface to drain within the porous overlay rather than on the surface of the pavement. PFC also provides water quality benefits. In this report, PFC and open-graded friction course (OGFC) are synonymous. Note that PFC and OGFC are not a full-depth permeable pavement and do not infiltrate runoff to the subgrade. Rather, runoff travels laterally through the overlay to the shoulder area, where the overlay terminates. Table 2-9 compares the influent and effluent median con- centrations for the TX1 site to the categorical performance estimates for permeable pavement for the entire BMPDB. As shown in the table, the TX1 facility shows superior TSS, TKN, and total phosphorus removal as compared to the cate- gorical performance estimates for permeable pavement. The TX1 facility underperforms in dissolved and total copper removal as compared to the categorical estimates. PFC shows strong performance for TSS and particulate-bound pollutant removal due to shallow sedimentation and filtration/straining processes that occur as stormwater passes through the pores of the PFC material. The limited capacity for mitigating dis- solved pollutants is reflected in the poor dissolved copper removal performance. Pollutants may become attached to the PFC matrix by strain- ing, collision, and other processes. Material that accumulates in the pore spaces of PFC is difficult to transport and may be trapped permanently. On the surface of a conventionally paved road, splashing created by tires moving through standing water can transport even large particulate matter rapidly to the edge of pavement. However, water velocities within the pore spaces of the PFC are low and likely could only transport the smallest material (Eck et al., 2010). PFC can produce TSS reductions consistent with the removal rates expected from practices such as sand filters or bioretention systems (Barrett, 2003; Hsieh and Davis, 2005; Hunt et al., 2008). Concentrations of total metals from PFC were generally significantly reduced when compared to those of conventional pavement. 2.2.2.3 BMP Water Quantity Performance (Peak Rate and Volume Reduction) Technical Summary of Volume Reduction from BMPDB. In 2011, Geosyntec Consultants and Wright Water Engineers, TSS (mg/L) Dissolved Copper (µg/L) Total Copper (µg/L) TKN (mg/L) Ortho- phosphate as P (mg/L) Total Phosphorus (mg/L) AustinTX1PFC In Out In Out In Out In Out In Out In Out 121 8 5.24 8.7 28.4 11.4 1.06 0.8 N/A N/A 0.16 0.05 Categorical BMP Performance (Geosyntec Consultants and Wright Water Engineers, 2012) Permeable pavement 65.3 13.2 5.37 5.75 13.07 7.83 1.66 0.8 0.04 0.05 0.15 0.09 Notes: Bold values indicate effluent median is statistically significantly less than the influent median (alpha = 0.05). Italicized values indicate effluent median is statistically significantly greater than the influent median (alpha = 0.05). Table 2-9. Median concentrations of AustinTX1PFC to Geosyntec Consultants and Wright Water Engineers (2012) categorical BMP performance summaries.

22 Inc., performed an analysis of the BMPDB to specifically eval- uate volume reduction through selected post-construction BMPs. Volume reduction is an increasingly important issue in TMDL and NPDES permit compliance; however, very little data within the BMPDB address this aspect of performance since the focus tends to be on pollutant concentration reduc- tion. The analysis notes that when volume data were present in the database, they were often suspected to be unreliable. However, a small percentage of studies were identified as having produced reliable volume reduction data. A summary of relative volume reduction observed from these studies is shown in Table 2-10. Normally dry vegetated BMPs (filter strips, vegetated swales, bioretention, and grass-lined detention basins) appeared to have substantial potential for volume reduction on a long-term basis, on the order of 30% for filter strips and grass-lined deten- tion basins, 40% for grass swales, and greater than 50% for bio- retention with underdrains. They also were shown to provide better volume reduction for smaller storms, which tended to occur more frequently than larger storms (Geo syntec Consul- tants and Wright Water Engineers, Inc., 2011). Retention ponds, wetland basins, and channels did not appear to provide substantial volume reduction on average, and they were not recommended for projects intending to achieve appreciable volume reduction. The study did not pro- vide specific data for BMPs (such as bioretention) that might use impermeable liners, but it was speculated that volume reduction performance would be lower compared to unlined systems subject to identical conditions (Geosyntec Consul- tants and Wright Water Engineers, Inc., 2011). University of Maryland Study of Volume Loss within Lined Bioretention Systems. In 2003, the University of Maryland constructed a lined bioretention system for the specific purpose of identifying reduction impacts to peak rate and volume. The study produced data from 49 storm events. In 18% of those, the lined bioretention systems pro- duced no discharge volume. When discharge occurred, it was observed for prolonged periods of time—sometimes several days—such that outflow hydrographs would overlap multiple storm events. Typical peak flow reduction (flow rate rather than volume) observed was on the order of 44% to 63% of the inflow peak rates (Davis, 2008). 2.3 Current Asset Management, Inspection, and Maintenance Practices 2.3.1 Asset Management and Inspection Needs State DOTs have been devoting effort to inventorying per- manent stormwater treatment facilities (by inspection of facilities) and their locations to support asset management and maintenance programs. The task takes years to complete, even with consistent, diligent effort. Many performance problems (and associated repair costs) can be identified and addressed early through a regular inspection program. WERF describes the value of inspection and monitoring of BMPs at various stages of development (WERF, 2012, p. 429): • Inspection during the design and construction phase helps ensure proper design, construction techniques, and sedi- ment and erosion controls. • Inspections following the construction phase serve to inspect, track, and help ensure that controls continue to function properly. • Regular monitoring during operation not only ensures that maintenance activities are being carried out as speci- fied, but also identifies any areas of potential system failure. • Standard inspection procedures help assess the stability and function of stormwater controls. 2.3.1.1 Basic Inspection Data The U.S. EPA advises development of inspection checklists to help determine renovation and repair needs for storm water BMPs. EPA recommends inclusion of the following general items within BMP inspection checklists (U.S. EPA, 2012): 1. The BMP’s minimum performance expectations, 2. Design criteria, BMP Category No. of Monitoring Studies 25th Percentile Median 75th Percentile Average Vegetated strips 16 18% 34% 54% 38% Vegetated swales 13 35% 42% 65% 48% Bioretention w/underdrain 7 45% 57% 74% 61% Grass-lined detention basin 11 26% 33% 43% 33% Table 2-10. Relative observed volume reduction in BMPDB data set.

23 3. Structural specifications, 4. Date of initial operation, 5. Expected life span, and 6. Maintenance requirements for each BMP, to help the inspector determine if a BMP’s maintenance schedule is adequate or in need of revision. In addition, a checklist will help the inspector determine renovation or repair needs. The WERF suggests that general BMP assessment include (WERF, 2012, p. 430): 1. Site conditions, 2. Water quality performance, 3. Structural integrity, and 4. Overall function. 2.3.1.2 As-Built Drawings A database or geographic information system (GIS) inven- tory of stormwater BMP locations should include other descrip- tive data for each facility. Inspectors need to know where the controls are and what they should look like so they can be maintained as designed. As-built drawings offer a number of advantages for this purpose (WERF, 2012, p. 430). They: • Provide details on components of a control that require inspection, • Reference operation and maintenance needs in some cases, • Reduce the potential for confusion in the field, and • Allow the inspector to verify that all parts of the facility are functioning as designed. Inspectors often lack ready access to as-built drawings, but agencies are increasingly investing in this access. Where as- built drawings are not available, some DOT NPDES program managers are consulting with maintenance staff to develop basic recorded information. DOTs are also increasingly requiring contractors to provide as-built drawings in an easily storable format. For example, Colorado DOT now has a requirement on any new project that the contractor provide a surveyed, final as-built drawing in electronic format (Gay, 2012). WSDOT has new NPDES permit reporting requirements, indicating that the agency will “work with project offices to develop a procedure for ensuring field verified as-builts are provided to headquar- ters as part of the project closeout procedure” and that 10% of new projects will be audited annually to verify that all reported newly constructed stormwater facilities are entered into the post-construction stormwater facility database cor- rectly (Washington State Department of Ecology, 2012). 2.3.1.3 BMPs Inspection Personnel Responsibility for inspection of post-construction storm- water controls varies widely across states and agencies. For example: • Regulatory agencies often have resource constraints, and consequently, routine inspections do not occur with the frequency typically recommended. In these instances, much of the regulatory response is complaint-based. • Annual certification of performance by owner or profes- sional engineer may be used. This is an idea that is cur- rently being discussed in California for compliance with municipal NPDES permits. • Colorado DOT stormwater personnel are assessing each BMP annually. • Harris County, Texas, requires that a professional engi- neer selected by the facility owner certify annually that all required maintenance for a given control has been per- formed and that the facility is functioning properly. • Maintenance staff or regional environmental staff per- forms level-of-service (LOS) condition assessments for features of stormwater facilities and other roadside assets in many states. • Contract staff/consultants perform inventories in other states, such as Delaware. 2.3.1.4 Recording and Storing Inspection Results for Performance Assessment While the results of inspections used to be stored on paper, they are increasingly recorded electronically for instant uploading to databases. Life-cycle performance assessment requires detailed attribute data that describe each feature’s material, defects, and repairs over time so that the reasons for failure can be understood. GIS combined with the hydraulic infrastructure database opens up a world of information about waterways, land use, and soil effects on the drainage system. At the Min- nesota DOT, drainage system and water quality features are captured by Global Positioning System (GPS) field inspec- tion or GIS tools and are accessible in a database called “HydInfra” (hydraulic infrastructure; http://www.dot.state. mn.us/bridge/hydraulics/hydinfra.html). Minnesota DOT’s web-based reports identify drainage system features that need cleaning or repair. Specialized reports simplify the end- of-year MS4 reporting requirements that need maintenance (see Figure 2-1). The HydInfra database includes inventory, inspection, and maintenance data on ponds, structural pollution control devices, MS4 outfalls, illicit discharges, pipes (culverts <10-in. span or storm drain pipes), structures (manholes, catch basins),

24 various special structures (aprons, end sections, weirs), and ditches (see Figure 2-2). Colorado DOT’s System for Recording Post- Construction BMP Assessments. Colorado DOT (CDOT) has been inspecting the full inventory of over 900 post- construction BMPs since 2010. Stormwater staff located and reviewed all BMPs in the field using information obtained from as-built plans and through consultation with maintenance staff. Stormwater staff record inspection results and reviews in the Stormwater Inspection Tool, a software application tailored to BMP types. Inspectors send results to maintenance staff to help identify labor/maintenance action needed to address identified issues. Maintenance performed and costs/labor hours are recorded in CDOT’s accounting database. CDOT annu- ally reports to the state regulatory authority on the number of post-construction water quality structures inspected, the total maintenance expenditures on each, and the results of limited, automated stormwater runoff monitoring. The state is currently developing a new online system to store BMP data. The system is being developed in C# pro- gramming language and SQL2005 and will be moved to CDOT’s virtual server as soon as it is ready (Gay, 2012). Maryland State Highway Authority’s Drainage Infra- structure Assessment System. The Maryland State High- way Authority (MDSHA) Drainage Infrastructure Assessment System was the first comprehensive system for recording and storing inspection results. MDSHA’s system was also the first evolved system to assess conditions in a tested, duplicable way (see Figure 2-3). MDSHA uses the system to manage the approximately 1,500 stormwater management facilities it owns, with inspec- tion teams of trained staff who identify potential further environmental improvements. MDSHA has complemented this work by mapping the entire state for opportunities for retrofitting BMPs, enhanced pollution prevention and stream restoration, and development of a plan for systematic imple- mentation of those improvements. The grade-based rating Figure 2-1. BMP information in GIS format, Minnesota Department of Transportation. Figure 2-2. HydInfra database information example.

25 system for stormwater management facilities includes an inventory, database, and photo record of all facilities state- wide and their maintenance status, within a GIS. Under the rating system, those installations graded “A” or “B” are con- sidered functionally adequate. By 2009, MDSHA had reached its long-term goal of 95% functional adequacy for its system, with that percentage being rated “A” = everything fine, work- ing fine, and no maintenance required or “B” = minor main- tenance (need mowing or trash removal), leaving only 5% needing maintenance or retrofitting to achieve functional adequacy. MDSHA’s drainage system GIS is designed to be used for planning-level computations and operations-level activities. The database is used to determine the general location of sys- tems and drainage areas, to track maintenance activities, and to address public complaints. Information in the drainage infrastructure database is intended to be sufficient to identify, locate, and evaluate every BMP to provide an overall assessment of MDSHA’s BMP inventory. The information in the system assists the agency with decisions on inspection, maintenance, repair, and retro- fit of BMP facilities, in addition to supporting compliance with MDSHA’s NPDES MS4 permit. It supports GIS queries: • By individual structure or system and BMPs (e.g., pipes, inlets, manholes, end walls, and their associated data attri- butes), • By outfall (size, type, etc.), • Within a drainage area, • Within a watershed, • Within a jurisdiction, • Statewide, and • By roadway contract. The system has evolved to also support hydrologic analy- sis of the drainage systems for the preparation of estimates of the quantity and quality of stormwater runoff from the SHA right-of-way and the effects of changes in stormwater management practices. More recently, MDSHA has added visual impact-assessment components to its evaluation and remediation. The managing for results (MFR) portion of MDSHA’s business and stewardship plan was used to measure the prog- ress and success of the NPDES program and define timelines and milestones for the numerous elements of the program. Using the MFR approach, MDSHA measured progress every month for each of the major elements and every 6 months for all the elements of the program. An example of this is the tracking of the required number of source identification efforts that needed to be completed. By tracking BMP facilities and progress, the database has also helped in identifying BMP failures. When MDSHA inspects an infiltration BMP, it does a functional rating and assesses whether the BMP is functioning. If a filtration struc- ture has failed, there may be an opportunity to convert the site into a structure of a different type—a wet retention facil- ity. MDSHA has had some success with converting failed infiltration BMPs to function as wet ponds. MDSHA is trying to assess how to efficiently reassign ratings. The database does have its limitations. While MDSHA’s database was developed to enable standardized, comparable, and meaningful data, MDSHA is finding that the agency does not have the staff and analytical resources to use the informa- tion to its maximum benefit. MDSHA’s system defines vari- ous filtering practices—for example, vegetated swales with Figure 2-3. MDSHA infrastructure system screen shot.

26 subcodes for wet, dry, and other swales—but the DOT main- tenance division is not able to use all of those subcodes in its record keeping or maintenance work. Delaware Department of Transportation (DelDOT) Stormwater BMP Inspection/Maintenance Program. In 2007, DelDOT, assisted by KCI Technologies, Inc., developed a statewide stormwater BMP inspection/maintenance pro- gram with a consistent protocol for inventorying, inspecting, and maintaining BMPs, now documented in DelDOT’s com- prehensive BMP Field Inspection Manual. Field tested in 2007 with the inspection of over 300 BMPs, DelDOT’s approach established four key components of a BMP inspection: site conditions, water quality, embankment, and outlet structure, each with specific evaluation parameters, differing among BMP types (Mattejat and Thompson, 2007). North Carolina LOS Rating and Performance Reporting for Post-Construction BMPs. The North Carolina Depart- ment of Transportation (NCDOT) has a system for evaluat- ing the LOS for post-construction BMPs. The LOS rating for stormwater control measures was created to establish a score for stormwater control measures being considered an asset to NCDOT and to gauge the maintenance needed for individual devices. A rating scale was developed from “A” to “F.” An “A” rating would be given to a device that shows some aging and wear but no structural deterioration or maintenance needs, and that is functioning properly. An “F” rating would be given to a device that is no longer functional due to the general or complete failure of a major structural component or the lack of adequate maintenance. Individual LOS ratings are taken at least once a year for all stormwater control measures. These ratings are averaged for divisions, counties, and road types and provided to the asset management group within NCDOT every 2 years. In addition, based on these average ratings, the division roadside environmental engineer (DREE) from each division is given a “does not meet,” “meets,” or “exceeds” rating that is found on his or her individual performance dashboard appraisal. Any rating below “C” indicates to the DREE that maintenance is needed on that particular device. NCDOT’s December 2010 Maintenance Condition Assess- ment Report shows over 94% of facilities functioning as designed, exceeding the 90% target the agency set for itself. With over 22,000 tenth-of-a-mile sample points, NCDOT has enough points to directly manage from its sample and confi- dently set maintenance budgets. Washington State DOT Maintenance Accountability Process (MAP) and LOS Rating. WSDOT has a MAP that uses outcome-based performance measures with a rating scale of “A” (best) to “F” (worst) for reporting the LOS pro- vided. Although WSDOT does complete turbidity and other monitoring for water quality, outcomes from the MAP do not necessarily refer to water quality measures. Rather, outcomes for WSDOT refer to tasks/results accomplished by mainte- nance personnel. This can be a percentage of proactive or preventive maintenance performed. WSDOT currently uses three types of assessments: opera- tional assessment, condition assessment, and task comple- tion. WSDOT has found task completion to be an important part of understanding what has and has not been done and whether budgets are sufficient. Operational assessment data indicate operational issues, such as how many repairs per sig- nal were needed in a given period. Conditional assessment data are collected using statistically valid, randomly chosen sites for field surveys. Task completion data are collected from records of work required and accomplished; this metric quan- tifies the number of tasks needed for a specific activity each year and how many of those tasks were completed. The tasks can be preventive maintenance with a scheduled frequency or can be a list of existing deficiencies. LOS is expressed as the percentage of identified tasks that were completed. The dif- ference between what should have been done and what was done identifies the backlog for individual maintenance activi- ties. Reporting using the task completion component began in 2010, with eight MAP activities. The 2011–2013 bienni- ums will expand the use of task completion to other MAP activities. The MAP priority matrix prioritizes maintenance activities and ranks them according to their contribution to maintenance program goals. 2.3.1.5 Drivers in Inventorying and Inspecting BMPs NPDES reporting requirements along with the additional urgency imposed by pollution reduction targets and TMDLs are driving increases in BMP inspection and maintenance to improve performance. For example, Rhode Island DOT’s NPDES permit requirements for pollution prevention/good housekeeping for municipal operations state that RIDOT must “develop inspection procedures and schedules for long-term operation and maintenance (O&M) of municipal facilities, municipal structural BMPs, and the MS4.” Asset management programs require basic data to allow decision makers to prioritize repair and budgeting for long-term O&M. DOTs need a record of installed BMPs, their mainte- nance requirements, and their maintenance history to ensure their operation at the design level. 2.3.2 Current Maintenance Practices 2.3.2.1 Current DOT BMP Maintenance Practices Determining Maintenance Frequency. Few DOTs have systematically or programmatically budgeted for mainte- nance of post-construction stormwater controls. When asked,

27 many DOTs indicated that maintenance is performed on an as-needed basis. Historically, the maintenance of stormwater BMPs included activities such as removing excess sediment, revegetating ditches and embankments, and trash removal that have occurred in response to inspection during a storm event. BMPs have also been maintained “on an emergency basis, when their hydraulic conveyance function is impaired enough to threaten the structural integrity of the highway or impair roadway safety” (WSDOT, 2005). DOT respon- dents reported that where formal information was not avail- able, maintenance guidelines for stormwater BMP guidance documents were based “mostly on regulatory judgment or historical estimates of sediment accumulation, rather than empirical data” (WSDOT, 2005). DelDOT defined remedial actions needed for each BMP after reviewing inspection results completed in 2007 and began to develop a long-term strategy for remedial actions, creating three general categories for remediation: mainte- nance work orders, invasive vegetation spray list, and retrofit recommendations. Sediment and vegetation buildup imped- ing the conveyance were the most common issues. For maintenance work orders, starting with the BMP inspections completed in 2008, routine maintenance issues identified in the inspections (performed by a third-party consultant) were entered into DelDOT’s maintenance work order system. Each DelDOT district assigned staff to receive BMP work orders and schedule tasks based on the type of work, location of the work, and severity of the issue. Work continues to be handled on a bulk rather than individual BMP basis for efficiency. For example, a labor crew might be scheduled to handle maintenance at several BMPs located in the same area, or a Vactor truck may be scheduled for removal of accumulated sediment from several BMPs. DelDOT is performing careful tracking of invasive species, including an inventory of the approximate square footage of various invasives at each BMP and an eradication strategy. For example, DelDOT eradicates Canadian thistle regardless of the amount observed. This tracking enables DelDOT to identify the needed level of funding and effort to address the issue in a timely way. Retrofit remedial actions are considered beyond the scope of DelDOT’s maintenance districts because these projects tend to require engineering analyses to redesign and recon- struct the BMP. DelDOT categorizes remedial actions as major (complete reconstruction) and minor (only a compo- nent of a BMP that needed repair or reconstruction). WSDOT has developed design standards for the basic BMPs used on its highway system. The standards are used for deter- mining what and when maintenance may be required, at given (typically annual) evaluation points. WSDOT has not estab- lished design standards for nonstandard BMPs, so the agency will start with literature values as they begin inspection and maintenance (Baroga, 2012). WSDOT identified the numbers of each BMP type in its inventory, by region, and maintenance requirements. For example, the agency computes that wet/ detention/infiltration ponds will need sediment clean out every 5 years, and it will take 3 days to remove 150 yd3 of sediment. CDOT’s approach assesses BMP function per plan specifi- cations in the field and then sends the field evaluation to the maintenance district/region for labor and equipment esti- mates. The evaluators “review as-builts, specs on how high the vegetation is supposed to be or what the sediment limits are, and work to restore the BMP to its intended function” (Gay, 2012). The Minnesota Department of Transportation (MnDOT) developed a BMP resource maintenance guide (Marti et al., 2009). The guide is a supplement to the state’s Stormwater Manual for inspection and maintenance activities for BMPs. It contains information for evaluating various BMPs to install based on anticipated long-term maintenance requirements. The University of Minnesota completed a study entitled “Assessment and Maintenance of Stormwater Best Manage- ment Practices” (Gulliver and Anderson, 2008). The docu- ment provides information on the assessment, maintenance, and renovation of stormwater BMPs. Rules of Thumb for Maintenance Schedule. DOTs and resource agencies often operate with rules of thumb regard- ing appropriate maintenance schedules. For example, the EPA’s stormwater pollution prevention fact sheet says that in storm- water ponds, “vegetation should be harvested every 3 to 5 years, and sediment removed every 7 to 10 years” (U.S. EPA, 2010). The New York State Department of Transportation (NYSDOT) has a GreenLITES (Leadership in Transportation Environmental Sustainability) program for maintenance and operations that indicates maintenance schedule cycle times on the following rotating schedule: • Every 10 years for open drainage facilities—maintaining ditches, shoulder grading, • Annual sweeping around closed conduit drainage, • Drainage structure repair on a 10-year schedule for closed conduit drainage, and • Capital improvements of closed drainage on a 50-year basis (NYSDOT, n.d.). The program provides recognition for the degree to which districts place catch basin inserts and culvert/pipe replacements in order to incentivize staff. This is primarily an internal management program for NYSDOT to measure performance, recognize good practices, and identify where it needs to improve sustainability practices. Influence of New NPDES Requirements on Maintenance Frequency. New NPDES permits are beginning to specify

28 maintenance schedules for permanent/post-construction stormwater BMPs. For example, WSDOT’s NPDES permit requires annual inspection and maintenance of stormwater BMPs (Baroga, 2012). According to the permit, BMPs must attain explicit standards, which are outlined in Chapter 5, Sec- tion 5 of WSDOT’s Highway Runoff Manual (WSDOT, 2011). Standard BMP Maintenance Activities. Most types of structural or vegetated BMPs share commonality in terms of the basic required maintenance activities. These activities typically include: • Restoration of eroded areas at inlets, outlets, and slope embankments; • Removal of invasive or excess vegetation; • Response to burrowing or nesting wildlife; • Response to standing water conditions or prolonged ponding; • Removal of any obstruction to maintenance access; • Identification and elimination of elicit discharge or other unusual occurrences in the vicinity, such as vandalism; • Repair of structural deformation, cracking, corrosion, joint failure, or settlement; • Removal of flow obstruction or excessive sediment buildup; • Replacement of damaged signs, fences, or other intended barriers to pedestrians or animals; and • Replacement of damaged or nonfunctioning irrigation systems (not typically used at most installations). Practitioners may refer to the following sources for more detailed maintenance checklists and activities associated with post-construction BMPs: • Best Practices Handbook on Roadside Vegetation Man- agement, Minnesota DOT (http://www.lrrb.org/media/ reports/200019.pdf). • U.S. EPA BMP Inspection and Maintenance webpage (http:// cfpub.epa.gov/npdes/stormwater/menuofbmps/index. cfm?action=factsheet_results&view=specific&bmp=91). • NCDENR [North Carolina Department of Environment and Natural Resources] Stormwater BMP Manual (http://portal. ncdenr.org/c/document_library/get_file?uuid=7b297ecd- 955a-417e-a024-56639b068f54&groupId=38364). • Northern Virginia BMP Handbook (http://www.novaregion. org/DocumentCenter/Home/View/1679) • Santa Clara Valley Urban Runoff Pollution Prevention Program Sample BMP Inspection Checklist (http://www. scvurppp-w2k.com/bmp_om_forms.htm). • Newton, Kansas, BMP Inspection and Maintenance manual (http://www.newtonkansas.com/Modules/ShowDocument. aspx?documentid=622). • Southeast Michigan Council of Governments BMP Main- tenance Inspection Checklists (http://www.semcog.org/ uploadedfiles/Programs_and_Projects/Water/Stormwater/ LID/LID_Manual_appendixF.pdf). 2.3.2.2 DOT Perceptions Regarding the Challenges of BMP Maintenance—Considerations for Design In NCHRP Report 728 (Geosyntec Consultants et al., 2012), DOT practitioners shared their insights and lessons learned on the selection, design, and implementation of BMPs, especially in urban environments. Maintenance issues were among their greatest concerns and their most frequently mentioned topic: • Retention/infiltration systems endanger groundwater or weaken pavement subgrade (Utah DOT). • Ponding in sand filters can lead to increased mosquito breeding (DelDOT). • It often is quite difficult to construct access roads for main- tenance forces (Oregon DOT). • Cartridge filters require trained staff and vehicle jib cranes and safe access to adequately maintain (Oregon DOT). • Ultra-urban BMPs require constant monitoring (New Mexico DOT and CDOT). • Accessibility for inspection and maintenance is often not considered in facility design but is essential to its life cycle, particularly for underground storage and treatment facili- ties. Facilities that use vegetation are often not successful due to stress of pollutants, wetness, drought, or improper species selection in conflict with desire to use native species (MDSHA). • Standing water and vector breeding due to inadequate soil conditions (New Jersey DOT). • Availability of training for maintenance personnel (Mon- tana DOT and CDOT). 2.4 BMP Life-Cycle Costs 2.4.1 Life-Cycle Cost Factors for BMPs The Center for Watershed Protection compiled cost data from 100 retrofit projects in a 2007 document, which pro- vided guidance for estimating construction costs (Center for Watershed Protection, 2007). The data reflect all types of retrofit projects, although the center noted that cost data for highway retrofits are sparse. Retrofit is defined as a stand- alone project without other highway construction. New con- struction in this report encompasses new and reconstruction projects.

29 Life-cycle cost factors are described in WEF’s 2012 manual, Design of Urban Stormwater Controls, which reviews and summarizes unit construction activity costs from standard civil engineering price guides, develops costing models to facilitate generic stormwater control cost estimation, com- pares actual and predicted costs, and outlines many cost fac- tors, especially in BMP construction (Barrett and WEF, 2012, pp. 486–489). Size, distribution, and complexity of storm- water systems and controls also affect maintenance needs. Some of the cost factors are: 1. Stormwater controls can be built at much lower costs as part of a larger project rather than as stand-alone projects. Larger projects offer better economies of scale and do not have as large a fraction of total cost for mobilization and project initiation. It is more cost-effective to grade in extra basins or swales when a much larger development site is already being graded. Similarly, wet basins and dry basins generally have lower unit costs as facility size increases. 2. Most cost studies assume building on undeveloped land, but some retrofits are built into existing public land or easements, while others require land to be purchased and may have higher costs to get the water to drain to the facility. Many sites are not in optimal hydraulic locations due to constraints imposed by prior development. In general, the construction costs for highway BMP retrofits can be quite high (as much as 10 times more expensive than new construction) and are highly site-specific (Currier et al., 2001). 3. Regulatory requirements vary for water quality control volumes and flow rates and for structure components such as inflow structures, splitter boxes, and fencing. Specified structural components can be complex and costly or simple and inexpensive. 4. Public entities often face more requirements, bidding laws, and regulation, entailing more supervision and steps, which can raise costs. Public agencies also take on long- term maintenance of their own projects, leading to an interest in making sure the work is done right and is sustainable. 5. If an agency is able to site a facility where little grading or excavation is needed and where blasting or long-distance hauling can be avoided, that generates savings, but many projects are subject to rules and regulations that limit the DOT’s ability to choose a more cost-effective site. 6. Some agencies have begun to seek partnerships with other entities (e.g., private developers or other agencies) to build stormwater controls with a better economy of scale and thus reduced cost. 7. Experienced staff and contractors are familiar with the steps involved and can suggest better, more main- tainable, and cost-effective designs and projects. New requirements and technologies are relatively costly since contractors have little experience with them. Likewise, inexperienced agency staff may not be confident or knowledgeable enough to suggest cost-reducing changes in rules and designs. 8. The number of bids can depend on the state of the econ- omy and the timing of the bid offering. If timed with many competing offerings, fewer bidders may respond, raising costs. 9. More stringent treatment requirements increase project costs. Water quality design criteria vary by jurisdiction and determine the size and complexity of the BMP that is required to meet them. 10. Geography and climate influence the design rainfall and rainfall-runoff characteristics of a site, in turn affecting drainage system component sizing. 11. The cost of land (purchase and legal costs) can outweigh design and construction costs for some controls in dense urban settings, making maintenance-intensive under- ground facilities seem more practical. Careful design and use of open space allocations sometimes reduces the effective cost of land allocated to surface water drainage. 12. Soil type and groundwater vulnerability dictate whether infiltration is required to treat an initial volume of run- off or whether additional storage and attenuation will be required. The soil type also dictates the level of erosion protection and vegetated reinforcement required and may influence plant selection. 13. Many stormwater control components require granular fill as the attenuation and filtering media; these costs will vary depending on the distance of the site from a poten- tial source. Topsoil costs will also depend on source loca- tions. Other market factors such as fuel costs to transport materials may greatly alter costs. 14. The availability of suitable plants and the required level of planting planned for a particular control component influence landscape costs, which can be substantial. In addition, landscape contractors are often required to provide a warranty for the plantings for some period, which can escalate with mortality rates of 20% to 25% for plantings. 15. Routine maintenance consists of basic tasks performed on a frequent and predictable schedule. These include inspections, vegetation management, and minor debris removal. In addition, three levels of routine maintenance can be identified, and these relate mainly to frequency of the activity being undertaken (and in WEF’s estimates of life-cycle costs). These are defined as: a. Low/minimum—A basic level of maintenance required to maintain the function of the stormwater control;

30 b. Medium—The normal level of maintenance to address function and appearance; it allows for additional activ- ities, including preventative actions, at some facili- ties; and c. High—Enhanced maintenance activities required for appearance and amenity only. 16. Intermittent maintenance typically consists of more heavy- duty, unpredictable, and infrequent tasks to keep systems in working order, such as repair of structural and erosion damage, and, potentially, complete facility reconstruc- tion. The intermittent category can include a wide range of tasks that might be required to address maintenance issues at a BMP (e.g., invasive species removal, animal bur- row removal, and forebay cleanout). Intermittent mainte- nance is nonscheduled, occurring as needed in response to field conditions. 17. Common maintenance activities are inspections, vegeta- tion management, and sediment removal, with frequency and thoroughness that can be affected by funding. Barrett et al. estimated that as much as 80% of total staff hours spent in the field in many jurisdictions is associ- ated with vegetated mowing, with little effect on near- term performance, as opposed to sediment, debris and trash removal, or structural repair, though lack of rou- tine maintenance can destroy structures in some cases, as when tree roots destroy embankments—a situation that can be avoided with periodic mowing (Barrett and WEF, 2012, p. 431). Common maintenance usually follows a regular schedule, or can if funding is available. When other contributions are leveraged, some costs can be heavily reduced or nearly eliminated. For example, highways are designed to provide a clear recovery zone adjacent to the road- way to enable drivers to regain control before they hit a fixed object or roll over. These roadside vegetated areas are built with low slopes up to about 10-m (30-ft) wide, precisely the design criteria to optimize the water quality benefits of vegetated strips, although these benefits have only recently been recognized (Barrett and WEF, 2012, p. 502). While DOTs made these land and design investments for transportation and safety purposes, they also provide water quality benefits. WEF concludes: “For swales and filter strips, water quality benefits can effectively be considered free when compared to conventional drainage sys- tems and when the maintenance is performed by the property owner” (Barrett and WEF, 2012, p. 509). Mowing is performed for safety/visibility and aesthetic purposes, but this is compat- ible with water quality objectives. The order of BMPs in a treatment train can also greatly affect maintenance costs and can produce substantial ben- efits when the last facility in a train, such as filters or infiltra- tion trenches or basins, can clog and require more expensive maintenance or rehabilitation. Again, swales or buffer strips offer an important benefit by reducing sediment upstream of a BMP that is more difficult to maintain. 2.4.2 Tracking Actual BMP Maintenance Costs Traditionally, DOTs have made only very rough estimates of the maintenance needs and costs of roadside assets, but now they are inventorying assets, creating asset registries, and establishing and tracking costs per unit to maintain and oper- ate those assets. A small number of DOTs are beginning to collect informa- tion on the true, real-time costs of maintaining stormwater controls. This involves assigning maintenance codes to struc- tures, individually identifying and attributing maintenance actions to individual BMPs located via GPS or automatic vehicle location technology, and creating the data systems and hiring staff to use them to perform the desired analyses. This information will provide the basic input data needed for finer-scale understanding and calculation of long-term performance and life-cycle costs of post-construction storm- water controls. DOTs can then follow the same process for full cost determination of permanent BMPs as for any main- tenance asset, as outlined in NCHRP Report 688: Determining Highway Maintenance Costs (Cambridge Systematics, Inc., et al., 2011): Step 1: Gather and classify maintenance program activities and expenditures. Step 2: Allocate maintenance support expenditures to line activities. Step 3: Gather and classify enterprise programs and expenditures. Step 4: Allocate a portion of enterprise support expendi- tures to the maintenance program. Step 5: Combine cost categories to derive full cost. WSDOT developed Excel spreadsheets with assumptions on maintenance needs, which were distributed to other states in the course of project interviews. WSDOT is estimating the costs for delivering the new BMP maintenance requirements in the agency’s latest NPDES permit. The NPDES permit requirements set out a clear regimen of design standards, from which WSDOT has been calculating costs. Cost projec- tions may not be needed for BMP maintenance in the future because the agency is within months of having maintenance vehicles fully GPS capable and able to report location, activ- ity, and hours spent for maintenance work, as well as removal quantities, such as the amount of sediment removed and cost. WSDOT staff and budget analysts anticipate that this will give the agency a better understanding of the costs of BMP maintenance (Baroga, 2012).

31 In WSDOT’s case, GPS data will be linked with the agency’s Highway Asset Tracking System (HATS), a tool for manag- ing maintenance activities by asset or roadway section. The system connects to highway features where the asset infor- mation of the agency is stored (existing asset ID, name, and location). Maintenance technicians document their work using a personal digital assistant (PDA) while simultaneously building/maintaining the inventory in HATS. When doing an inspection on an asset, staff will have the capability to add the asset or generate a pending activity, recording deficiencies that require action to be taken. The action could be anything from making a specific repair, cleaning, or making a recom- mendation for a larger repair. The system will track when, where, and what was inspected; if a pending activity was gen- erated from the inspection; when the pending activity was completed; and if it remains to be completed. The system will also track multiple work activities within a section of roadway and create a pending repair for those items that cannot be completed at the time. 2.4.3 Historic Data and Studies Relating to BMP Life-Cycle Cost 2.4.3.1 Caltrans Retrofit Study The Caltrans retrofit study included detailed accounting of BMP capital and maintenance costs, which were also sub- jected to independent third-party review (Caltrans, 2004). The final report points to uncertainty with regard to the location-specific nature of some costs and to how well the cost data may reflect actual costs in a large-scale retrofit pro- gram. However, the data are detailed and comprehensive, and can provide a means for comparing and ranking costs asso- ciated with various BMP technologies. Tables 2-11 and 2-12 BMP Type Cost/m3 of the Design Storm ($) Delaware sand filter 3,472 Multichambered treatment train 847 Wet basin 2,670 Oil–water separator 2,540 Austin sand filter 2,009 Infiltration trench 1,954 Storm filter 1,575 Swales 951 Unlined extended detention basin 877 Strips 835 Infiltration basins 639 Lined extended detention basin 348 Continuous deflective separator 220 Drain inlet inserts 33 Table 2-11. Actual construction cost of BMP technologies (1999 dollars). BMP Equipment and Materials ($) Average Labor Hours Sand filters 872 157 Extended detention basin 958 188 Wet basin 2,148 485 Infiltration basin 3,126 238 Infiltration trench 723 98 Biofiltration swales 2,236 246 Biofiltration strips 1,864 233 Storm filter 308 106 Multichambered treatment train 2,812 299 Drain inlet inserts 563 121 Oil–water separator 1,066 139 Continuous deflective separator 785 254 Table 2-12. BMP actual annual maintenance effort for Caltrans BMP Retrofit Pilot Program.

32 provide capital and maintenance information from the Caltrans study. The practitioner may refer to the following website for other detailed BMP capital and maintenance cost informa- tion from the Caltrans retrofit study: http://www.dot.ca.gov/ hq/oppd/stormwtr/Studies/BMP-Retro-fit-Report.pdf. 2.4.3.2 Highlights from WEF Life-Cycle Cost Analyses of BMPs WEF and WERF have produced life-cycle cost analyses for a variety of BMP types (Lampe et al., 2005; Barrett and WEF, 2012, pp. 502–509). Unit whole life costs are provided in Table 2-13. Maintenance costs of wet basins make up almost 50% of the whole life cost when basins are implemented in high-visibility locations, where aesthetics are at a premium. Dry basins tend to be easier and less expensive because there is little or no standing water in the facility. Wet and dry basins cost the same to construct if there is no pond liner for the wet basin. The primary maintenance cost of bioretention is associated with vegetation management. The frequency of this activity was assumed to be similar to swales but with a greater cost because many bioretention facilities would require weeding, mulch replacement, and other activities beyond the mowing required for most swales. For swales and filter strips, water quality benefits can effec- tively be considered free when compared to conventional drainage systems, as well as when the maintenance is per- formed by the property owner. Infiltration trenches may require little routine mainte- nance outside of litter and debris removal. The whole life cost driver is the frequency with which the trench must be reha- bilitated. Intervals of 4, 8, and 12 years were assumed based on low, medium, and high scenarios, at which time the cost is essentially the same as the original construction cost. For infiltration basins, the capital cost and routine maintenance are essentially the same as those for a dry basin, but an infil- Stormwater Control Whole Life Cost ($/m3) Low Maintenance Medium Maintenance High Maintenance Swales/strips 500 660 2,200 Wet ponds/wetlands 520 600 925 Dry extended detention basins 330 375 575 Sand filter 450 520 670 Bioretention 1,900 2,200 5,100 Infiltration trench 1,200 1,600 2,700 Infiltration basin 330 400 700 Permeable pavement 570 640 1,400 Table 2-13. Whole life costs of common BMPs per cubic meter of stormwater treated (WERF, 2012). tration basin can incur much higher costs associated with maintaining sufficient infiltration rates. In addition to sedi- ment removal, an infiltration basin may require additional activities to remove and replace clogged soils on the floor of the basin. The frequency of this activity is largely dependent on the initial soil texture and the rate at which sediment accu- mulates in the basin. With PFC pavement in the same location as a conventional surface, the cost for the water quality control facility is the incremental cost difference between a conventional pavement and the pervious overlay pavement. The difference in whole life cost depends on the frequency of sweeping. DOT inter- est has been fostered through safety and livability co-benefits offered: better visibility and traction in storm events, reduced splash and hydroplaning, and reductions in deflected noise from highway traffic. Porous asphalt overlays (PFCs) are being used in Georgia, California, Massachusetts, and Utah. PFC was up to 8.1% of all pavements in Texas in 2010. The pavement is assumed to need replacement more frequently (every 25 years or less versus 35 and 40 years) at a cost equal to original con- struction. Water quality monitoring of three locations in the Austin area indicated up to a 90% reduction in pollutant dis- charges from PFC compared to conventional pavement. This reduction is the result of accumulation of pollutants within the pavement and the reduction in pollutants washed off vehicles during storm events (Eck et al., 2010). 2.4.3.3 Urban Denver Drainage and Flood Control District BMP-REALCOST Tool—A Predictive Tool to Estimate BMP Life-Cycle Costs The Urban Denver Drainage and Flood Control Dis- trict (UDFCD) BMP-REALCOST tool produces order-of- magnitude cost approximations for use primarily at the planning level: • Construction costs are estimated using a parametric equa- tion that relates costs to a physical parameter of a BMP:

33 total storage volume (for storage-based BMPs), peak flow capacity (for flow-based or conveyance BMPs), or surface area (for permeable pavements). • Maintenance costs are estimated using a derived equation that relates average annual costs to a physical parameter of the BMP. • The additional costs of designing and permitting a new BMP are estimated as a percentage of the total construction costs. For Denver-area projects, a value of 40% is recommended if no other information is available. • The cost of purchasing land for a BMP is estimated using a derived equation that incorporates the number of impervi- ous acres draining to the BMP and the land use designation in which the BMP will be constructed. • The costs of administering a stormwater management pro- gram are estimated as a percentage of the average annual maintenance costs of a BMP. For Denver-area projects, a value of 12% is recommended if no other information is available. • After some period in operation, a BMP will require major rehabilitation. The costs of these activities (including any salvage costs or value) are estimated as a percentage of the original construction costs and applied near the end of the facility’s design life. The percentages and design lives vary according to BMP (UDFCD, 2010, pp. 2–17). UDFCD’s BMP-REALCOST tool produces net present value (NPV) of the whole life costs of the BMP(s) imple- mented, the average annual mass of pollutant removed (lb/year), and the average annual volume of surface runoff reduced (ft/year), which can then be used to compute a unit cost per pound of pollutant or cubic feet of runoff removed over the economic life (years) of the BMP (UDFCD, 2010, pp. 2–17). 2.4.3.4 Research Funded or Supported by the Minnesota Department of Transportation MnDOT has participated in or is aware of the development of recent studies on the cost, maintenance, and assessment of BMPs. Two of these projects are described in the following. “The Cost and Effectiveness of Stormwater Management Practices” (Weiss et al., 2005). Stormwater management prac- tices for treating urban rainwater runoff were evaluated for cost and effectiveness in removing suspended sediments and phosphorus. Construction and annual operating and mainte- nance cost data were collected and analyzed for dry detention basins, wet basins, sand filters, constructed wetlands, bioreten- tion filters, infiltration trenches, and swales using literature that reported on existing sites with stormwater management practices across the United States. The annual operating and maintenance costs were also compiled. “Best Management Practices Construction Costs, Main- tenance Costs, and Land Requirements” (Barr Engineering Company, 2011). This report summarizes a typical range of low-impact development stormwater management BMP costs and identifies a range of construction and operating costs for eight treatment low-impact development BMP cate- gories. The costs and the expected longevity of the BMPs were used to estimate life-cycle costs for these stormwater BMPs.