Field Test of BMPs Using Granulated Ferric Oxide Media to Remove Dissolved Metals in Roadway Stormwater Runoff (2019)

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.

5 Chapter 2 Field Testing of Ferric Oxide Media Filters: Project Design 2.1 Field Testing Sites and Tributary Watersheds Two existing ferric oxide filters were chosen for this study. Both sites have similar design characteristics: (1) filtration media-a mixture of sand and ferric oxide (parent material is elemental iron), (2) downflow filtration, (3) stormwater storage above the cell, (4) multiple inlets that deliver water to the top of the cell, and (5) perforated underdrains located at the bottom of the sand-ferric oxide media that feed a single outlet. The Woodlynn Avenue site (see Figure 2-1) is located in the Maplewood Mall, Maplewood, Minnesota, and receives runoff from a 100 percent impervious parking lot for which 1.81 acres are tributary to the treatment cell. The Highway 36/61 site (Figure 2-2) is located in a cloverleaf of Highway 36 in Maplewood, Minnesota and receives direct highway runoff (AADT in 2017 of 68,000 for Highway 36, source: Minnesota Department of Transportation) as well as mixed commercial runoff from adjacent areas (total watershed area of 29.5 acres). The Woodlynn Avenue ferric oxide filter was constructed in 2011 with the first full year of operation being 2012. The Highway 36/61 ferric oxide filter was constructed in 2013 with the first full year of operation in 2014. The use of ferric oxide media in a sand filter matrix is a relatively new technology for stormwater treatment and the number of existing treatment cells with the desired attributes was somewhat limited. Although the Woodlynn Avenue site does not receive highway runoff, the design is analogous to a swale with a check dam for storage and this type of design could be readily applied to the highway right of way. The Highway 36/61 ferric oxide filter was designed and constructed as part of a highway reconstruction and stormwater treatment project and was a logical site for study. Figure 2-1 Woodlynn Avenue swale-type ferric oxide filter during and after construction.

6 Figure 2-2 Highway 36/61 vault-type ferric oxide filter during and after construction. From aerial and ground-based photographs of the Woodlynn Avenue and the Highway 36/61 cells, differences are apparent in the tributary watersheds and this influenced designed decisions, affects runoff chemistry, and site hydrology and hydraulics. The tributary watershed to the Woodlynn Avenue filter is 100 percent impervious and directly tributary to the filter. Storm even runoff peaks quickly requiring that the filter provide storage (maximum potential storage area of 6,839 ft3) and then gradual treatment by filtration to capture meaningful runoff volume. The filter size was also dictated by space availability. The imperviousness of the Woodlynn Avenue watershed implies that runoff here may be more dominated by inorganic solids, lower alkalinity, hardness, and dissolved solids. The land use of the watershed tributary to the Highway 36/61 site is open space developed (7.7 percent), low intensity developed (6.8), medium intensity developed (53.8 percent), and high intensity developed (31.6 percent), with an overall estimated imperviousness of 63 percent (Hommer et. al., 2012). The watershed incudes some depression storage in swales, intermittent wetlands, and open space. The inflow hydrograph is longer at that Highway 36/61 filter compared to the Woodlynn Avenue filter. There is also greater potential for solids removal in upstream depressions and ponds at Highway 36/61. The size of the filter bed at Highway 36/61 (22,800 ft2) was based upon the watershed size but also the ample space at this location enabled the construction of a larger filter. 2.1.1 Ferric Oxide Filter Design 2.1.1.1 Swale-Type Filter at Maplewood Mall The Woodlynn Avenue ferric oxide filter consists of two inlets (Woodlynn North and Woodlynn South) and a bypass inlet that flows when the basin is full and overflowing (Figure 2-3). There is one outlet fed by a perforated underdrain below a ferric oxide filter bed consisting of a mixture of sand and ferric oxide (parent material is elemental iron which clearly rusted to ferric oxide prior to the cell being placed online, see Figure 2-1). The outlet pipes (two total) are restricted by caps with 2-inch diameter holes on each drain pipe. The area of the cell is 3,903 ft2 at an elevation of 940.5 feet, which is also the overflow elevation. According to the design drawing plan set, the sand-ferric oxide filter bed is 160-feet long with a total estimated area of 528 ft2. This length and width was confirmed with field measurements. The design drawings also indicate that the depth of the ferric oxide sand media was planned to be approximately 2.5 feet. Field measurements indicate that the sand-ferric oxide filter media bed is approximately 1.3-feet deep to the top of the underdrain. The sides of the treatment cell were covered with approximately 1 foot of planting soil covered with 3 inches of wood chips.

7 Figure 2-3 Modified design drawing for the Woodlynn Avenue ferric oxide-sand filter. 2.1.1.2 Vault-Type Filter at Highway 36/61 The ferric oxide filter at Highway 36/61 consists of two inlets with the HWY Inlet West inlet fed by direct highway runoff and runoff to the HWY Inlet Pond originating from highways and commercial properties as well as some open space consisting of swales and small intermittent wetlands (see Figure 2-4). The Highway Inlet East delivers runoff to a pre-treatment pond which in turn discharges to the ferric oxide filter at the HWY Inlet Pond location. The pond is designed to settle solids. There is a geotextile barrier and soil berm between to the pond and the filter and it is likely that minimal water filters through this berm to the filter. The filter bed consists of a mixture of sand and initially elemental iron (recycled scrap material often called zero valent iron) which clearly had rusted to a significant degree (see the red color of the sand mixture in Figure 2-2) to ferric oxide prior to the filter being placed online. The sand and ferric oxide filter bed was designed to be approximately 0.8 feet deep with the total bed depth of 1.6 feet the inclusion of pea gravel on the bottom of the surface of the bed. Field measurements indicate that the ferric oxide-sand bed depth is 1.2 feet. Below the pea gravel is a geotextile fabric that was designed to control solids mobility rather than prevent infiltration. Designs called for perforated underdrains with side slits to be placed in a 0.6-feet deep bed of pea gravel. The perforated underdrains feed to a header that collects all of the filtered and treated water. According to field measurements, there is an overflow outlet set at 0.6 feet above the filter bed.

8 Figure 2-4 Modified design drawing of the Highway 36/61 ferric oxide filter. 2.1.1.3 Differences of the Ferric Oxide Filter Systems Studied There are several significant differences between the designs of these two systems; the differences and how they are expressed in the monitoring data are discussed in Chapter 3. Storage and size of the ferric oxide-sand filter bed area are probably the most notable. The Woodlynn Avenue treatment cell provides 6,389 cubic feet of storage and it often takes several hours after a storm event for the cell to drain. The watershed (78,844 square feet) to filter bed area (528 square feet) ratio is 149. For the Highway 36/61 cell, only 0.6 feet of storage is allowed before water enters the overflow outlet and water only infrequently ponds above the filter bed. The watershed (956,412 square feet) to filter bed area (22,800 square feet) ratio is 42. At the Woodlynn Avenue cell approximately 1 foot of planting soil was placed adjacent to the ferric oxide filter bed, covered with woodchips, and planted with vegetation (see Figure 2-3). The filter bed at Highway 36/61 does not have plantings or organic matter that are inundated with stormwater and the surface of the bed is covered with pea gravel. The Highway 36/61 treatment site also has a pre-treatment pond that removes a large percentage of stormwater solids. 2.1.1.4 Ferric Oxide Media Samples of the in-place ferric oxide sand media were collected in September, 2016 to identify the elemental composition and mineralogy of the samples. The elemental composition was analyzed at the University of Minnesota Department Of Earth Sciences using energy dispersive X-ray spectroscopy (EDS) and scanning electron microscopy (SEM). The results of these analysis are summarized in Table 2-1. Although the design target for each treatment cell was 5 percent iron aggregate (elemental iron or zero valent iron) by weight for both of the ferric oxide sand filters studied, it appears that the fraction of iron by weight in the ferric oxide sand filter bed was notably higher. This does not impact this study’s findings. The concentration of several metals (as weight percent) are higher in the Woodlynn Avenue filter bed

9 material compared the Highway 36/61 material. Also provided for comparison is the elemental analysis of the iron aggregate used as the parent material for the ferric oxide filter (Table 2-2). These data were provided by the iron aggregate supplier. Table 2-1 Elemental analysis of ferric oxide-sand filter bed samples using dispersive X-ray spectroscopy and scanning electron microscopy. Element  Weight (%)  Woodlynn  Avenue  Highway 36/61  O  55.22  67.17  Na  0.97  0.00  Mg  0.96  0.83  Al  4.61  2.79  Si  20.89  20.04  S   0.36  0.00  K  1.24  0.35  Ca  0.66  0.34  Ti  0.10  0.00  Mn  0.17  0.00  Fe  14.83  8.50  Table 2-2 Elemental composition of iron aggregate as provided by the supplier. Element  Iron Aggregate Weight %  Element  Iron Aggregate  Weight %  Fe  87‐93  Mo  0.08‐0.15  C  2.85‐3.23  Ti  0.004‐0.1  Mn  0.14‐0.60  Cu  0.11‐0.20  S  0.67‐0.107  Al  0‐0.005  P  0‐0.132  Co  non‐detect  Si  1.0‐1.85  Mg  0.01  Ni  0.05‐0.21  B  0.01  Cr  0.03‐0.23  Zn  0.01  V  non‐detect  Zr  0.01  Samples of the ferric oxide-sand filter bed were also collected in June 2018 to further understand the minerology of the iron in these filters using Magnetic Properties Measurement System zero-field-cooled and field-cooled series magnetometry. The analysis was conducted at the University of Minnesota as part of a study on the minerology of ferric oxide in Minnesota iron enhanced sand filters (personal communication, Beth Fisher). Preliminary analysis indicates that the primary iron oxide species in these ferric oxide-sand filters is goethite, maghematite, magnetite, and hematite.

10 2.2 Monitoring Design 2.2.1 Objectives and Sampling Design The overall objective of this study is to monitor two full scale ferric oxide-sand filters to quantify the capacity of a ferric oxide media used in a filter bed to remove dissolved metals from natural stormwater runoff and under natural hydrologic conditions. The monitoring design was scoped to enable the practitioner to estimate potential metals removal as a percentage and load reduction for filtration practices that use ferric oxide in a similar manner to the systems evaluated in this current study. Collection of data necessary to support design recommendations and life cycle cost estimates was also a main objective. Several task level objectives are listed below:  Water balance: Collect data to develop a detailed water balance for each storm event such that the mass of metals loaded to the ferric oxide-sand filter can be accurately compared to the mass of metals leaving the system through the treatment cell outlet. This task consists of three sub- tasks below. o Flow: Continuously measure surface water inflows to and outflows from the treatment systems such that the complete storm event hydrograph could be used to accurately calculate the water balance and hence event mean constituent concentrations at the inlet and outlet of the system. o Level: Install level sensors in each ferric oxide-filter to the degree necessary such that the average level of water in the filter bed can be determined for each storm event. o Rain gauge: Install, calibrate, and properly maintain a rain gauge at the Highway 36/61 site with a range of 5 inches per hour, an accuracy of 1 percent at rates up to 1 inch per hour and precision of 0.01-inch increments. An existing rain gauge is maintained near the Woodlynn Avenue site.  Low level metals: To minimize the collection of metals data that are below reporting limits, analyze dissolved and total metals to reporting limits that are below 1 part per billion for most metals, except for iron which is 6.0 µg L-1 and zinc which is 3.0 µg L-1.  In-situ chemistry: Collect pH, dissolved oxygen, and temperature data at one location within the ferric oxide-sand filter bed to enable evaluation of the effect of these parameters on treatment performance. Maintain and calibrate these probes to the frequency needed such that the data are available for each storm event.  Recordkeeping: Use best practices with respect to recordkeeping (field notebook, chains-of- custody (COC)), maintain records during the course of the project, and use methods that minimize error. All field data will be maintained in a commercial database and direct data deliveries from

11 the analytical laboratories will be maintained in Barr’s Laboratory Management Information System to minimize data management error. 2.2.1.1 Monitoring Locations Woodlynn Avenue Figure 2-3 shows the sampling layout of the Woodlynn Avenue site. Flows were measured using area velocity meters at four locations: (1) North Inlet, (2) South Inlet, (3) Overflow Inlet, and (4) Outlet. Water sampling was conducted at three locations: (1) North Inlet, (2) South Inlet, and (3) Outlet. The volume of the Woodlynn Avenue filter as a function of the depth above the filter bed was calculated from a detailed survey of the site conducted in 2017. Pressure transducers placed in the filter bed provided estimates of water depth in the filter bed as well as the depth of water above the filter. Highway 36/61 Flow monitoring was conducted using a combination of weirs and pressure transducers or area velocity flow meters for in-pipe flow measurement. Figure 2-4 shows the layout of the Highway 36/61 site. Flows were measured at three locations: (1) east inlet manhole (HWY East Inlet) facilitated by a compound weir installed in 2017 for this project, and (2) west inlet manhole (HWY West Inlet) facilitated by a 90 degree V-notched weir was installed in 2017 for this project, and (3) outlet (HWY Outlet). Water sampling was conducted at four locations: (1) east inlet manhole (HWY East Inlet), (2) west inlet manhole (HWY West Inlet), (3) outlet of the pre-treatment basin which discharges to the filter bed (HWY Pond), and (4) the outlet (HWY Outlet). Water monitoring at HWY Pond was triggered by flows at the HWY East Inlet manhole. 2.2.1.2 Monitoring Parameters Monitoring parameters were chosen based upon the (1) frequency of occurrence in highway runoff and how often given parameters are measured (Granato and Jones, 2019, Barrett et al., 2014), or (2) the parameter was an indicator of the treatment system function, or (3) the parameter had the potential to modify the function of the ferric oxide filter (Smith, 1999, Cismasu et al., 2016). Both total and dissolved metals were measured as it was expected that particulate metals removal data would be useful as part of treatment performance evaluation. Table 2-3 provides a summary of the parameters monitored, methods, and reporting limits.

12 Table 2-3 Parameters monitored, method and reporting units. Test Method  Analyte  Reporting  Limit  Units  Metals  EPA 1638 Mod  As  0.080  µg/L  EPA 1638 Mod  Cr  0.15  µg/L  EPA 1638 Mod  Cu  0.30  µg/L  EPA 1638 Mod  Fe  6.0  µg/L  EPA 1638 Mod  Pb  0.030  µg/L  EPA 1638 Mod  Ni  0.24  µg/L  EPA 1638 Mod  Zn  3.0  µg/L  General Parameters  SM2540 D‐97  TSS  0.6  mg/L  SM2540 E‐97  VSS  0.6  mg/L  SM5310C‐00  DOC  0.5  mg/L  SM2320 B‐97  Alkalinity, Total (as CaCO3)  10  mg/L  SM2320 B‐97  Alkalinity, Bicarbonate (as CaCO3)  10  mg/L  SM4500‐Cl E‐97  Cl  1  mg/L  SM2340 C‐97  Total Hardness (as CaCO3)  5  mg/L  2.2.2 Equipment and Materials Field equipment deployed for this study is discussed below. All of the field probes were checked for accuracy or were calibrated in accordance with the required frequency for each probe. The light dissolved oxygen sensor and pH sensor were calibrated at the start of the season and the calibration was checked weekly. The pH sensor was recalibrated periodically during the season. Pressure transducers were checked for accuracy at the start of the season and evaluated after data download which occurred after each storm event that yielded samples. A general inspection was conducted with each site visit and may include clearing debris or sediment obstructing area velocity meters, hoses and weirs, assessment of the integrity of cables, sampling equipment, and battery charge. 2.2.2.1 Highway 36/61 Field equipment deployed included:  Five pressure transducers (HOBO U20L-004) for placement in the filter bed  Four autosamplers with batteries  An area velocity meter and three Campbell Scientific pressure transducers (Campbell CS451) for use with the weirs for flow measurement  Two weirs  pH probe (Campbell Scientific CS526 pH Probe)

13  Dissolved oxygen probe (HOBO DO Data Logger U26-001)  Tipping bucket rain gauge (Texas Electronics 525-L Rain Gage with 6-in. Orifice)  CR1000 Campbell datalogger with battery  Weather-proof enclosures with locks  Solar panel  RAVENXTV Cellular modem with antenna (telemetry unit) 2.2.2.2 Woodlynn Avenue Field equipment deployed included:  Three Isco autosamplers  Four 2150 Isco AV Flow Meter Module  Weather-proof enclosures with locks  Three pressure transducers (HOBO U20L-004)  pH probe (Campbell Scientific CS526 pH Probe)  Dissolved oxygen probe (HOBO DO Data Logger U26-001) 2.2.3 Quality Assurance and Quality Control Data quality objectives (DQOs) are qualitative and quantitative statements that clarify project objectives, define the appropriate type of data, and specify tolerable levels of potential decision errors that will be used as the basis for establishing the quality and quantity of data needed to support decisions. The DQOs for the project were used to develop and implement procedures for field sampling, chain-of-custody and note taking, laboratory analysis and electronically gathered field measurements which include level, pH, and dissolved oxygen, and reporting that will provide the level of data required for determining the characteristics of monitored stormwater (Barr Engineering, 2017). Measurement quality objectives (MQOs) are statements that support the project DQOs and contain specific units of measure that are directly compared to data. The purpose of this section is to address the MQOs for data quality indicators (precision, accuracy, representativeness, and sensitivity), along with the means by which they are measured to monitor compliance with the project needs. Precision Precision measures the reproducibility of a measurement under a given set of conditions. Precision of field sampling was assessed by comparing the analytical results between field duplicate samples. Field duplicate samples were submitted to the laboratory as blind (masked) samples. The relative percent difference (RPD) was calculated for each pair of duplicate analyses where both results are greater than five times the reporting limit (RL). The normal and duplicate samples were reported as an average of the two values. Field duplicate samples were collected and sent to the laboratory for each sampling event alternating between the Highway 36/61 and Woodlynn Avenue sites. RPDs ≤ 30% will be considered acceptable (when both the native and field duplicate sample concentrations are greater than five times the RL). RPDs above 30% for water samples will result in corrective actions or qualification by the QA/QC Lead.

14 Accuracy For the analytical results, accuracy is evaluated as the degree of agreement between an observed value and an accepted reference value and measures bias in a measurement system. Accuracy will be addressed by calibrating field and laboratory instruments, and by analyzing laboratory quality control samples. The recovery limits for accuracy are expressed in terms of acceptable percent recovery of a known quantity. Representativeness Representativeness is a qualitative parameter that is defined as the degree to which data accurately and precisely represents a characteristic of a population, a parameter variation at a sampling point, a process condition, or an environmental condition. Representativeness was assessed on this project through the use of equipment blanks and field duplicate samples. The equipment blank and field duplicate samples were collected at a frequency of one per sampling event alternating between the study locations. The equipment blank was a vigorous quality control as the laboratory supplied blank water was placed in the autosampler bottle, collected in the same manner as the samples, placed in the sample splitter and collected from the splitter, and filtered if dissolved samples were collected. Potential contamination during sample collection and processing would be identified as part of this procedure. Sensitivity Sensitivity expresses the methodology’s and laboratory’s ability to meet the reporting limits as shown in Table 2-3. Completeness Completeness is a measure of the amount of valid data obtained from a measurement system compared to the amount that was expected to be obtained under normal conditions. Rejected data, or sampling points that do not yield a usable sample count against the percent completeness. Field completeness goals for this project is greater than 90 percent. It is expected that the laboratories will provide data meeting QC acceptance criteria for 90 percent or more of all samples tested. Following completion of analytical testing, completeness was calculated as the ratio of valid data divided by the number of targeted data. Completeness for this study was 96 percent. Comparability Comparability is defined as the confidence with which one set of data can be compared with another. The extent to which existing and planned analytical data will be comparable depends on the similarity of sampling methods, sample preparative procedures, analytical methods and holding times. Comparability will be satisfied by ensuring that proper and consistent sampling techniques are followed. This will be accomplished by the project team and measured with the use of quality control samples as well as adherence to the laboratory and field SOPs. The sampling and analytical techniques were not altered during the two year study except for a change in pH measurement at Woodlynn which was conducted in the laboratory rather than in-situ.