Article pubs.acs.org/est
Biofilters for Stormwater Harvesting: Understanding the Treatment Performance of Key Metals That Pose a Risk for Water Use Wenjun Feng, Belinda E. Hatt,* David T. McCarthy, Tim D. Fletcher,† and Ana Deletic Department of Civil Engineering, Centre for Water Sensitive Cities, Monash University, Building 60, VIC, 3800, Australia S Supporting Information *
ABSTRACT: A large-scale stormwater biofilter column study was conducted to evaluate the impact of design configurations and operating conditions on metal removal for stormwater harvesting and protection of aquatic ecosystems. The following factors were tested over 8 months of operation: vegetation selection (plant species), filter media type, filter media depth, inflow volume (loading rate), and inflow pollutant concentrations. Operational time was also integrated to evaluate treatment performance over time. Vegetation and filter type were found to be significant factors for treatment of metals. A larger filter media depth resulted in increased outflow concentrations of iron, aluminum, chromium, zinc, and lead, likely due to leaching and mobilization of metals within the media. Treatment of all metals except aluminum and iron was generally satisfactory with respect to drinking water quality standards, while all metals met standards for irrigation. However, it was shown that biofilters could be optimized for removal of iron to meet the required drinking water standards. Biofilters were generally shown to be resilient to variations in operating conditions and demonstrated satisfactory removal of metals for stormwaterharvesting purposes.
■
impact on taste and color in water.8 In addition, Fe is of particular concern to aquifer storage and recovery schemes because of its potential to block injection and recovery wells. The design of stormwater biofilters to date has consequently been optimized for removal of pollutants of concern to environmental protection but not for water supply applications. Removal of Cu is often in excess of 80% from stormwater3,4,7,9,10 and outflow concentrations of Zn, Pb, and Cr are generally below the Australian guidelines for environmental protection.9 However, the processes that govern this removal, and which are crucial for optimizing the design of the biofilters, are not well understood. Since previous research has shown that metals are largely captured in the top 50−100 mm of the filter media,11 it follows that filter media depth will not be a primary determinant of metal removal in biofilters; however, a minimum depth of 300 mm is required to support plant growth.12 However, this has not been tested for more than a handful of metals, so it is unclear whether more reactive metals will follow this pattern. Read et al.13 found that vegetation was not important for metal removal (statistical differences in outflow concentrations between species were observed, but the small variations between plant types had no practical significance), although this could be because their study was
INTRODUCTION The rapid rate of urbanization and climate change experienced during recent decades has created challenges for conventional urban water management.1 Both the quantity and quality of natural water resources have been depleted; thus, nontraditional water sources such as wastewater recycling and stormwater harvesting are becoming increasingly popular. In particular, stormwater, which is usually of better quality than wastewater, has excellent potential as a water resource due to a higher level of public acceptance and large volumes being generated near to where the water is needed.1,2 Despite this, stormwater harvesting as a practical method has not been widely adopted to date, in part due to a lack of treatment technologies that can affordably deliver fit-for-purpose water.2 One of the most promising stormwater treatment systems is biofiltration, a technology that has demonstrated good performance for reducing polluted stormwater discharges with the objective of protecting receiving waters; numerous studies report on the removal efficiency of biofilters for sediment, nutrients, and metals, which are all of concern to the health of receiving waters, e.g., refs 3−6. However, studies addressing the capacity of biofilters to remove pollutants that are important with respect to human use of the treated stormwater, such as pathogens and toxic chemicals, are limited. Although treatment of metals of direct relevance to aquatic health (usually Zn, Cu, Cd, and Pb) has been studied,3−7 no work has been done on the behavior of iron in biofilters, which is of high concern for stormwater-harvesting schemes due to its highly detectable © 2012 American Chemical Society
Received: Revised: Accepted: Published: 5100
September 27, 2011 February 9, 2012 March 8, 2012 April 12, 2012 dx.doi.org/10.1021/es203396f | Environ. Sci. Technol. 2012, 46, 5100−5108
Environmental Science & Technology
Article
Read et al.,13 Carex appressa, Melaleuca ericifolia, Microlaena stipoides, Dianella revoluta, and Leucophyta brownii were the selected plant species; nonvegetated (soil only) columns were used as controls. Sandy loam (with a d50 of 0.25 mm) was used as the primary filter material for all configurations, with different additives: (i) vermiculite and perlite, both exploded minerals with a high cation exchange capacity (hypothesized to have long-term benefits for metal removal10), and (ii) organic matter, which showed a good performance for metal removal (see Table S2 in the Supporting Information for soil chemical characteristics). Filter media depths of 300, 500, and 700 mm were selected to represent the practical range of depths; smaller depths are insufficient to support plant growth, while larger depths would lead to considerable head losses that are rarely allowed in urban drainage systems. Three different inflow volumes (50, 25, and 12.5 L) were used to test different sizes of biofilters, representing 1%, 2%, and 4% of the catchment area, respectively; these sizes were based on typical current practice and design guidelines in Australia. The biofilter columns were dosed twice per week continuously from June 2006 to April 2007, allowing almost a year of both climatic vegetation and seasonal variations in vegetation growth. The dosing volumes were equivalent to a rainfall depth of 4.53 mm, that is, a 1 in 4 month average recurrence interval (ARI) event (5 min duration) or a 1 in 2−3 month ARI event (10 min duration) for Melbourne,18 the maximum storm size that biofilters are typically designed to capture and treat. This also reflects a typical Mediterranean climate pattern. Semisynthetic stormwater was prepared following the procedure described previously,7,9,11 by mixing dechlorinated tap water with sediment collected from a stormwater pond. After sampling this slurry to determine concentrations of the pollutants of interest, laboratory-grade chemicals were added to top up concentrations to typical event mean concentrations (EMCs), based on reviews of Australian and worldwide stormwater quality19,20 (see Table 1). The metals were therefore mainly in dissolved form to mimic operationally challenging conditions. The choice to use semisynthetic stormwater rather than natural stormwater was based on the need to ensure consistency of concentration and composition, so as to avoid unintended artifacts. Given the volumes required, the supply of stormwater, harvested and delivered to the laboratory, would not have allowed us to achieve constant concentrations. We used a constant-mixing tank to mix the stormwater prior to its application to the columns. To ensure uniform input concentrations between each of the columns, the designated volume of stormwater was appliedusing jugsin a series of 5 passes so that each column received the same proportion of water from the dosing/mixing tank as it was drawn down. Inflow pH was measured for the first five sampling campaigns. Eighteen samples were taken for each campaign (based on three samples per batch and six batches being needed to dose all the columns). The average pH value (±1 standard deviation) was neutral (6.0 ± 0.5, n = 100), although a pH of less than 6.3 was measured in 13% of the samples. This pH is consistent with the typical pH of urban stormwater.19 In this experiment, the water used for dosing also had a similar hardness, in the range of 15−30 mg/L,21 to values in Australian stormwater.22 The dosing water had a raw alkalinity of around 11.5 mg/L; Makepeace23 suggests a range of 8−1273 in worldwide stormwater quality.
conducted over a very short period (only several months). Finally, and most importantly, the impact of clogging on metal removal needs further investigation. A field experiment conducted by Le Coustumer et al.14 showed that around 43% of biofilters in Australia have hydraulic conductivities lower than that specified by Australian biofilter design guidelines. Their study also showed that this clogging effect was governed by the top layer of filter media, meaning that removal performance through the filtration media is unlikely to be affected, but overall removal performance may be affected, due to more frequent bypass. Vegetation is important in controlling hydraulic conductivity in biofilters10 and could influence the pH and other chemical conditions of soils.15,16 Therefore, an examination of the interplay between key design characteristics (e.g., vegetation and soil type and filter depth) and its influence on metal removal over a longer time frame is required. Finally, it is also important to understand the influence of operational factors, such as the effect of inflow volume and concentration, on treatment of metals. This paper focuses on the treatment performance of stormwater biofilters with respect to removal of Fe, Cu, Cr, Zn, Pb, and Al, all of which are important to the utilization of harvested stormwater and environmental protection. It examines the impact of key design parameters (vegetation, filter media type, and filter media depth) and operating conditions (inflow concentration and volume) and interactions between these factors on metal removal. Two hypotheses were initially formed on the basis of the aforementioned studies: (i) filter media type is an important design characteristic, while vegetation type and filter media depth are not, and (ii) neither inflow pollutant concentration nor inflow volume should be a significant operating factor in the metal removal process. However, both of these hypotheses were found to be only partially true, with significant differences between the metals.
■
MATERIALS AND METHODS Experimental Setup and Procedure. A total of 120 biofilter columns were constructed from 375 mm diameter PVC pipe with a depth to hold 300−700 mm of filter media and a 210 mm sand and gravel drainage layer. A 400 mm transparent Perspex top section allowed for plant growth (allowing natural light conditions) and ponding of water. We used construction methods consistent with typical construction practice of biofilters; the sandy loam was thus only given light hand-compaction upon addition with regular inundation by water for a week immediately after construction to achieve hydraulic compaction of the media. The large (375 mm) columns were chosen in order to provide the most realistic representation of a full-scale biofilter. While the column diameter greatly exceeded the minimum column:particle diameter ratio of 50 recommended to avoid wall effects due to preferential flow down the smooth internal walls,17 large columns were considered necessary to allow root development representative of real systems. To assess the influence of design parameters and operating conditions on metals removal, we tested five different plant species, three different filter media types and three different filter media depths, as well as three inflow volumes and two inflow pollutant concentrations. Five replicates were tested for each configuration. It can be seen in the detailed configurations shown in the Supporting Information (Table S1) that we did not have a fully factorial experimental design; this was due to high experimental costs. On the basis of findings from 5101
dx.doi.org/10.1021/es203396f | Environ. Sci. Technol. 2012, 46, 5100−5108
5102
concn
volume
type
none C. appressa D. revoluta M. stipoides L. brownii M. ericifolia C. appressa C. appressa M. stipoides M. stipoides M. ericifolia M. ericifolia C. appressa C. appressa C. appressa C. appressa M. stipoides M. ericifolia M. ericifolia M. stipoides C. appressa
std std std std std std std std std std std std std std low high high high std std std
700 700 700 700 700 700 500 300 500 300 500 300 700 700 700 700 700 700 700 700 700
media typed
IDg IDg
freshwater ecosystem protection28
93 (2) 94 (2) 92 (4) 93 (3) 93 (5) 81 (8) 92 (5) 93 (2) 95 (1) 92 (4) 90 (1) 86 (7) 92 (4) 93 (2) 94 (4) 92 (2) 91 (3) 81 (6) 87 (8) 96 (2) 97 (1)
removal (%)
Cu
b
0.005 (5) 0.005(11) 0.003(29) 0.005(10) 0.004(41) 0.005(24)
