of a ten year old stormwater biofilter in Sweden
Sara Eklund
Natural Resources Engineering, master's 2020
Luleå University of Technology
Department of Civil, Environmental and Natural Resources Engineering
An evaluation of the treatment performance of a ten year old stormwater biofilter in Sweden
Sara Eklund
Department of Civil, Environmental and Natural Resources Engineering Luleå University of Technology
I
II
Acknowledgements
First I would like to thank my supervisor Senior Lecturer Godecke-Tobias Blecken at the Department of Civil, Environmental and Natural Resources Engineering, Luleå Technical University, Sweden.
A big special thanks to Lian Lundy, professor of Environmental Science at Middlesex University in Great Britain. Without you Lian there would not have been a thesis.
I would like to show my gratitude to the employees at the Tekniska kontoret in Tyresö for their help during all my visits. A special thanks to Krister “Krille” for his patience when helping me in Tyresö and answering a million questions.
There are many others that have been both directly and indirectly involved in this project. I would like to thank all of you. This thesis would never have been written without all the support I received from you. Amongst these people I would especially like to mention Kerstin Nordqvist, Gesche Reumann, Helen Galfi and Anna Magnusson at the Department of Civil, Environmental and Natural Resources Engineering, Luleå Technical University, Sweden. I would also like to mention Niclas Bergqvist at MJK Automation. Thank you for all the help and support you have given me.
Last but not least I would like to thank my family and friends, you are the best!
III
IV Abstract
Urban runoff may be contaminated with, for example, metals, nutrients and sediment.
Generally, such runoff enters waterways and oceans without any type of treatment.
Bioretention systems can be used to protect the aquatic environment since one of their main objectives is to remove pollutants from stormwater. This study presents data on one of Sweden´s first biofilters, constructed 10 years ago in the municipality of Tyresö. The main aim of this thesis is to evaluate the treatment performance of an operational biofilter through a comparison of inlet and outlet stormwater quality. The objectives include:
• Determine the concentrations of selected heavy metals and nutrients in stormwater at the inlet and outlet of the biofilter
• Identify if, and if so how, heavy metal concentrations and nutrient concentrations change after treatment
• Assess the condition of the biofilter in relation to particle size distribution and constitution of the filter media compared to data generated in an earlier study
Ten sampling campaigns were conducted in 2015 involving the collection of stormwater at the inlet and outlet. Rainfall data was also collected for each event. Samples were analysed for selected metals (lead, cadmium, copper, zinc, nickel and chromium) and nutrients (nitrogen and phosphorous). At the end of the sampling campaign, samples of the biofilter’s filter media were analysed for particle size distribution. Data from a previous study conducted in 2013 was also used in this thesis to provide a baseline data set for comparison. The event mean concentrations of most monitored substances were lower in the outflow than inflow, except for chromium which typically showed an increase. Calculated removal efficiencies show considerable variation, with low removal efficiencies in comparison to previous studies reported in the literature. The low removal performance may be explained by the relatively low pollutant concentrations in both inflowing and outflowing stormwater. If the inflow concentrations are close to the “irreducible concentrations” (Cirr) of a stormwater facility, no further reductions are likely. There could even be negative removal as even small variations in concentration can generate (translate into) substantial negative increases. Comparison of the current data set with the results of the study conducted in 2013 indicate that the treatment performance of the biofilter has not changed suggesting that, despite the colder climate, biofilters can continue to remove pollutants from stormwater after a ten year period of operation. Analyses showed that the filter media has the same particle size distribution as when the biofilter was constructed. Deeper samples of the filter media have lower metal concentrations than those reported in samples collected at the surface of the biofilter.
Keywords: biofilter, water quality, heavy metals, nutrients, stormwater treatment, Sweden, Tyresö
V
VI
Sammanfattning
Urbant dagvatten kan innehålla olika föroreningar, till exempel metaller, sediment och näringsämnen. Vanligtvis når dagvattnet vattenvägar och hav utan någon rening alls. Gröna dagvattensystem kan användas för att skydda den akvatiska miljön då en av deras huvuduppgifter är att rena dagvatten. Den här rapporten presenterar data från Sveriges första biofilter som byggdes för 10 år sedan i Tyresö. Huvudmålet med den här rapporten är att utvärdera reningsförmågan av Sveriges första biofilter. Fokus ligger på att jämföra vattenkvalitén från inflödet med vattenkvalitén från utflödet. De frågeställningar som utreds är:
• Fastställa koncentrationen av utvalda tungmetaller och näringsämnen i dagvattnet vid inflödet och utflödet på biofiltret
• Identifiera om och hur tungmetallernas och näringsämnenas koncentration ändras efter infiltration
• Utvärdera biofiltrets status med tanke på filter-materialets innehåll och partikelstorlek jämfört med tidigare studie
Under 2015 utfördes tio provtagningar då dagvatten samlades in både från inflödet och utflödet. Regndata samlades också in för varje regnhändelse. Proven analyserades för utvalda metaller (bly, kadmium, koppar, zink, nickel och krom) samt näringsämnen (kväve och fosfor). I slutet av provtagningskampanjen samlades även filtermaterial från biofiltret in och analyserades för fördelning av partikelstorlek. Resultat från en tidigare studie utförd under 2013 har också använts i den här studien som jämförande data. Medelkoncentrationen för varje regnhändelse (EMC) hos de flesta föroreningarna var lägre i utflödet än i inflödet, dock gäller det inte för krom, som istället visade på en ökning. Jämfört med tidigare utförda studier visar resultaten från den här studien på både lägre och mer varierande avskild mängd föroreningar. Den låga reningseffekten kan förklaras med att både inkommande och utgående föroreningshalter är låga. Om inkommande koncentrationer är i samma storleksordning som minsta utgående koncentration eller ”oreducerbar koncentration” (Cirr) tappar dagvattenanläggningen sin förmåga att rena dagvattnet. Reduktionen av föroreningar bli till och med bli negativ om inkommande koncentrationer är lika med eller längre än utgående koncentrationer. Resultaten från jämförelsen mellan studien från 2013 och studien från 2015 visar att biofiltrets reningskapacitet inte har ändrats med tanke på föroreningskoncentrationer.
Det indikerar att biofilter fortfarande kan ha kapacitet att rena dagvatten även efter att ha använts i tio år. Analyser visade att filtermaterialet har samma partikelstorleksfördelning som när biofiltret byggdes. Prov tagna djupt ner i biofiltret visar lägre föroreningskoncentrationer än prov tagna vid biofiltrets yta.
Nyckelord: biofilter, vattenkvalitet, tungmetaller, näringsämnen, dagvattenrening, Sverige, Tyresö
VII
VIII Contents
1 Introduction ... 10
1.1 Background ... 10
1.2 Objectives ... 11
2 Literature review ... 12
2.1 Stormwater ... 12
2.2 Stormwater pollution ... 12
2.2.1 Heavy metals: sources, mobility and toxicology ... 13
2.2.2 Nutrients ... 15
2.2.3 Road salt ... 16
2.2.4 Total suspended solids ... 16
2.3 Stormwater biofilters ... 16
3 Method and materials ... 20
3.1 Site description ... 20
3.2 The structure of Öringevägen (Tyresö) biofilter ... 21
3.3 Equipment setup ... 21
3.4 Sampling ... 22
3.4.1 Water sampling ... 22
3.4.2 Filter media sampling ... 23
3.5 Analysis ... 23
3.5.1 Water samples ... 23
3.5.2 Filter media analysis ... 24
3.6 Data Analysis ... 24
4 Results ... 26
4.1 Water quality ... 26
4.2 Comparison between studies conducted in 2013 and 2015 ... 29
4.3 Filter media metal concentrations and particle size distribution ... 30
5 Discussion ... 32
5.1 Water Sampling ... 32
5.2 Comparison of data collected during the 2013 and 2015 sampling campaigns ... 34
6 Conclusions ... 36
6.1 Recommendations for future research ... 36
7 References ... 38
8 Appendices ... 42
8a. Additional water quality data ... 42
8b. Flowlink data ... 50
9
10 1 Introduction
Earth has gone through a rapid urbanization during the last century. In 2007 the urban population exceeded the rural population for the first time and in 2014, 54 % of the planet´s population was reported to be urban-dwelling-. This trend is expected to continue and by 2050 the world´s population is predicted to be 66 % urban (United Nations, 2014). In Sweden, levels of urbanization are even more pronounced, where 85 % of the population is urban (Svanström, 2015). Growing cities entail both advantages and disadvantages. For example, growing cities offer opportunities to many people such as employment and education. On the other hand, urbanization increases the demand on environmental resources and volumes of waste increase, both traits that can lead to sever environmental and health problems (Moström, 2013).
As cities grow the pressure on societies’ infrastructure increases. One of these infrastructure services is stormwater management. The urbanization and densification of Swedish cities has led to a fragmentation and decrease in green areas and an increase of hard impermeable areas. This has the effect of a decreased amount of permeable area in the urban landscape, which in turn means there is less area where stormwater can infiltrate into the ground (Naturvårdsverket, 2015). The natural water cycle is also changing due to the effects of rapid climate change. For example, precipitation is expected to increase, especially during winter, in Sweden. Also, the frequency of intense high volume rains is expected to increase. This change in precipitation will lead to a higher flood risk around lakes and along waterways as well as an increase in the risk of erosion and landslides. Sea level is also expected to rise, which will impose a flood risk in low coastal areas. This will affect infrastructure, communication and housing areas (SMHI, 2015). Another issue that needs to be addressed is the degradation of water quality. Increases in the number of heavy precipitation events will increase the amount of runoff entering lakes and waterways, transporting with it metals, sediments, nutrients, trash, animal waste and other pollutants. This can cause problems for the water infrastructure and water supply (Svenskt Vatten, 2007). This increases the requirement for a long-term functional stormwater management strategy (Naturvårdsverket, 2008).
1.1 Background
In Sweden, the dominant system for stormwater management up until the beginning of the 1950´s was a combined sewage system which carried sewage, rain, snowmelt and – depending on the catchment and state of piped system – groundwater. As urban areas grow, the volume of water generated during rainfall and snow melt events can lead to the exceedance of capacity, with combined systems potentially posing a high risk of flooding cellars and other low points during intense rains. Combined sewer overflows cause environmental pollution of receiving waters. During the 1960s, a system that managed sewage water and stormwater in separate pipes was implemented (Svenskt Vatten, 2007). This directed stormwater and snowmelt direct to the closest receiving water. However, pollutant loads within these flows can, on occasion, be comparable to loads associated with sewage and hence an alternative solution was sought. About 20 years ago a new approach to managing stormwater was introduced called sustainable stormwater management (Stahre, 2004). The idea of the new concept was to create a system that mimics nature´s way of managing stormwater by looking to infiltrate surface water flows, and where this was no possible, to retain flows for release at a slower rate once the event had passed. A range of these sustainable drainage systems (SuDS) are available including swales, green roofs, infiltration basins, biofilters, wetlands and ponds. As well as mitigating stormwater volumes, these methods can also address water quality issues as a result of their promotion of a range of biological and physicochemical processes e.g. sedimentation, microbial degradation and photolysis (Stahre, 2004). To maintain and create such green-blue surfaces is increasingly
11
seen as an important part of the urbanization process and something that has to be considered early when planning new areas.
The internationally recognized concepts Water Sensitive Urban Design (WSUD) and Low Impact Development (LID) (design strategies which promote the use of systems such as SuDS) may be implemented to minimize the negative impacts of the changing “natural” water cycle, offering benefits to both public health and the aquatic environment. WSUD and LID are alternatives to traditional planning strategies as they use a more holistic approach and consider ways in which infrastructure and planned stormwater systems may be integrated with a site´s natural features. WSUD and LID can incorporate a range of SuDS (also referred to as stormwater Best Management Practices; BMPs) in the US (Moreton Bay Waterways and Catchments Partnership, 2006). One widely used type of SuDS are biofilters which aim to retain stormwater flow and improve its quality using a range of physical, chemical and biological processes including filtration by soil media and uptake by vegetation. Biofilters and their treatment efficiency have been studied within a temperate climate, with little consideration to their usability in colder climates (Blecken, 2010). This study aims to provide a foundation for further development of biofilters and understanding of their treatment performance in colder climates.
1.2 Objectives
The main aim of this thesis is to evaluate the water quality treatment performance of Sweden´s first biofilter situated in Tyresö. The research questions to answer include:
• What are the concentrations of heavy metals and nutrients in the stormwater at the inlet and outlet of a 10 year old biofilter?
• Do the heavy metal and nutrient concentrations change after treatment?
• What is the condition of the biofilter regarding the particle size distribution and constitution of the filter media compared to a previous study?
These research questions were addressed though completion of the following objectives:
• Collection of inlet and outlet water samples during ten rainfall events
• Analysis of samples for a range of metals (lead, cadmium, copper, zinc, nickel and chromium) and nutrients (nitrogen and phosphorous)
• Collection and analysis of filter material samples at different depths and from different locations within the biofilter
• Analysis of data to understand system performance
• Development of recommendations to support the development of sustainable plans for stormwater management that will benefit both the aquatic and urban environment
12 2 Literature review
2.1 Stormwater
Precipitation that has fallen on and passes over the impermeable surfaces associated with an urban area is defined by Butler and Davis (2004) as surface runoff. Surface runoff can mobilise and transport a range of pollutants which can cause negative impacts within receiving waters and may also cause flooding (Butler & Davis, 2004). Water environments are critical assets to society as, for example, sources of drinking water and due to their environmental, aesthetic and recreational values. Natural environments are often transformed by urban construction, with particular impacts on the water environment. Any surface changes within a catchment area will alter the water quantity and quality (Ashantha, et al., 2005). Impervious areas increase the volume surface runoff (since less water can infiltrate the ground or evapotranspirate) and the speed at which it flows. This has the effect of increasing the total water volume reaching a river or stream during or soon after a rain event, with runoff discharging at high velocities from a single discharge point. The increase of hard impervious areas hence increases the lag time of the hydrograph and the peak flow, see Figure 1. Thus, urbanization imposes a greater risk of flooding as well as impacting receiving water quality (Butler & Davis, 2004).
Figure 1. Effects of urbanization on a runoff hydrograph (Butler & Davis, 2004).
2.2 Stormwater pollution
Stormwater may contain a wide range of substances originating from a diversity of sources including the atmosphere, vehicles, buildings, roads, animals, de-icing activities, urban debris, spills and leaks (Butler & Davis, 2004). Pollutants that have been reported in stormwater include sediments, salts, oil, grease, nutrients, heavy metals and bacteria (Lundy et al., 2011).
Stormwater composition varies extremely, both between and during storm events. Parameters
13
affecting stormwater quality and quantity include catchment characteristics, land use, season, amount of precipitation and duration of the preceding dry period (Blecken, 2010).
2.2.1 Heavy metals: sources, mobility and toxicology
Sources Stormwater can carry many pollutants (see Figure 2 for an overview of sources, substances and pathways). Heavy metals are one important pollutant group given special focus in this study as the metals lead, cadmium, copper, zinc, nickel and chromium are analysed for in this study. These are the heavy metals most often reported in relation to stormwater pollution (Blecken, 2010).
Figure 2. Principal stormwater pollutant sources and types (Lundy, et al., 2011).
Lead is a common metal found in soils and rocks all over the world, albeit in low concentrations. The average concentration in the earth’s crust is 16 ppm (Åslund, 1994). Lead is primarily mined from the mineral galena (PbS), and it requires low temperatures to be extracted. Studies indicate that civilizations extracted lead from galena by burning wood 3500 years ago. Today lead is mostly used in batteries but earlier it was used as pigment in paints, as a drying agent in oil paints and as an agent in gasoline and in ammunition. Other environmental concerns are atmospheric emissions due to ore smelting, which is responsible for 80 % of anthropogenic emissions now its use in petrol has been banned. Natural sources of lead in the atmosphere are forest fires, volcanic eruptions and soil particles carried by the wind (Laws, 2000). In natural waters lead is mostly found as carbonate or chloride complexes.
Lead may also bind to organic compounds, clay minerals or suspended compounds.
Especially in very acidic soils, lead binds strongly to organic compounds. Due to the typical formation of low soluble compounds such as lead carbonate, lead phosphate, and lead sulphate, the mobility of lead is often restricted in soil environments (Åslund, 1994).
In natural waters cadmium is normally found as suspended complexes with iron and manganese or bound to organic compounds. There is a strong connection between pH and
14
the partitioning of cadmium between water and sediment compartments. For example, generally in natural waters, as the pH reduces the concentration of cadmium in solution increases. As noted earlier, key factors influencing cadmium mobility are pH and redox potential. These variables determine if cadmium is absorbed by clay minerals, iron oxides or sulphate oxides, which greatly reduces its mobility in soils. The ion Cd2+ is the dominant form of cadmium in surface waters when the pH is below 8 (Åslund, 1994). Weathering and erosion are natural sources of cadmium. However, the anthropogenic emissions are in general 2,6 times larger than natural sources. The major source is the manufacturing, use and disposal of end products, mainly nickel-cadmium batteries and emissions from traffic-related activities e.g.
cadmium-containing road paints. Mining activities are responsible for more than 50 % of atmospheric emissions (Laws, 2000).
Zinc is a common mineral in soils and rocks. It mostly forms hydroxy or carbonate complexes or is found as free zinc ions. In natural waters zinc is often bound to organic complexes and found in low concentrations. The solubility is closely connected to pH. At a pH below 5 the concentration of zinc in solution increases (Åslund, 1994). Zinc is widely used, for example, to galvanize iron and steel, as the negative electrode in electric dry cells and as a component of brass alloys (Ecyclopedia Britannica, 2016). A major source of zinc in urban environments is the wear and tear of car tyres and the degradation of street furniture. In natural waters high levels of zinc may be harmful to fish and other aquatic organisms (Tölgyessy, 1993).
Copper is a very common substance that has been used by humans for millennia. Today copper is widely used in both industry and agriculture. Therefore, increasing amounts of copper end up in the environment. Copper can be released to the environment both by natural and anthropogenic sources. For example, copper may be released by forest fires, sea spray, decaying vegetation, mining activities, and combustion of fossil fuels. In relation to traffic, a key source of copper is brake linings. Copper binds to organic matter or minerals in the sediment or soil. Therefore, copper is generally retained in surface soils. However, it can travel far in surface waters as free ions or in association with suspended particles.
Chromium does not occur freely in nature. It is mined as chromite (FeCr2O4) and discharged in surface waters via different industries and from traffic activities. The metal is mainly used in alloys, in metal ceramics and chrome plating. Whilst chromium (III) is an essential nutrient for organisms, chromium (VI) is hazardous to many organisms. Chromium enters the environment as chromium (III) with chromium (VI) formed through natural and anthropogenic activities such as emissions from traffic activities. Most of the chromium ends up in waters or soils where it strongly attaches to soil particles (Lenntech, 2016).
Organic matter has a strong ability to absorb nickel. Nickel is widely used in many different applications e.g. numerous alloys. For humans, nickel is essential in small quantities but becomes toxic to human health if the uptake is too high. Generally, nickel will adsorb to soil particles or sediment and become relatively immobile. However, in acidic soil nickel is more mobile and may therefore enter groundwaters. At elevated concentrations nickel is hazardous to organisms and animals. Nickel may inhibit growth of algae and cause cancer in animals living near nickel sources (Lenntech, 2016).
Mobility in surface waters Heavy metals are introduced to the environment in different ways such as weathering processes, volcanic eruptions or anthropogenic activities, e.g. traffic, industrial emissions mining and construction (Laws, 2000). The final sinks for non-volatile metals are often sediments and soils. However, if these substrates are disturbed and/or there is a change in physico-chemical conditions, previously bound metals can be released back into the environment. The mobility of heavy metals in the environment is dependent on the speciation of the metal. Speciation is controlled by physico-chemical conditions e.g. redox
15
potential and pH, and these parameters determine a metals’ environmental behaviour. Most heavy metals are insoluble in alkaline to neutral conditions. However, organisms may directly assimilate the metals e.g. as a result of ingesting particles which have adsorbed metals to their surfaces. Also, metals associated with particulate matter will remain within sediments (e.g. though formation of precipitates) until disturbed when the sediment sink effectively becomes a sediment source Several studies have found that the concentration of heavy metals is higher in sediments than in the water (Laws, 2000).
Toxicology A range of metals are found in the Earth’s crust. Some of them are vital to humans, plants, and animals, meaning that they function as trace elements (Laws, 2000). By definition, all metals are non-degradable and may therefore accumulate in receiving soils, biota and waters if the intake/input is higher than the excretion/output. Plants and animals may accumulate metals through a range of processes including uptake, ingestion and sorption enabling metals to enter the biological system. Concentrations of metals may then biomagnify (increase in concentration as they move up the food chain). However, it is not only the total concentration that makes metals toxic. Their potential toxicity also depends on the form and state of the metal. Freely dissolved ions are often more toxic than complex molecules. When short organic molecules bind to metals they become more soluble in contrast to binding to longer chain molecules and their bioavailability may increase as a result. This leads to easier transport of metals to and within an organism (Schnoor, 1996). The major reason for the toxicity of heavy metals is due to their high affinity for sulphur. Some heavy metals bind easily to amino acids containing sulphur, which may lead to enzyme or cell interference (Åslund, 1994). Not all heavy metals are acutely toxic in cases of direct exposure. Examples are manganese, zinc, copper, and iron. However, if exposed to these for a sufficient period of time chronic toxicity to aquatic organisms and humans may occur.
2.2.2 Nutrients
Nutrients are another relevant stormwater pollutant group (Blecken, 2010). Nutrients of interest are phosphorous and nitrogen as their discharge to receiving waters can lead to negative environmental impacts such as eutrophication.
In the atmosphere, the most common form of nitrogen is as nitrogen gas (N2). Due to the strong bonds between the nitrogen atoms, nitrogen gas is very stable. However, there are bacteria and algae that can bind nitrogen gas direct from the atmosphere and transform the gas into nitrates which are a form which can be taken up by plants and thereby introduced into biota (Stumm & Morgan, 1996). Whilst background concentrations of nitrate in groundwater are generally not at a concentration that can cause harm, the concentrations may increase due to anthropogenic activities such as the use of fertilizers in agriculture which can infiltrate into underlying groundwaters or runoff from fields into surface waters following rainfall events (Svensson, 2005). Nitrates support the growth of algae and aquatic plants, which in turn provide food for other aquatic organisms. However, when too much nitrogen enters the ecosystem the algae and aquatic plants grow too fast (known as a bloom). When these plants die, microbial populations explode in number within the receiving waters, rapidly breaking down the organic matter. This process rapidly uses up dissolved oxygen which may result in fish death and lethal impacts on other organisms due to the resulting oxygen depletion. Some algal blooms may also be toxic to humans and other organisms since they can produce toxins.
As with nitrogen, sources of phosphorous include fertilizers, manure and organic wastes.
Phosphorous typically occurs as phosphates which readily attach to soil particles and are transported by surface runoff. It is an essential nutrient but in high concentrations also causes algae and aquatic plants to bloom, which may lead to eutrophication and associated reduction in dissolved oxygen levels (USGS, 2015). Hence, phosphorous may cause fish death and disturb ecosystems in the same way as nitrogen (EPA, 2016). However, in freshwater bodies
16
phosphorous levels are often identified as the limiting nutrient for primary production (Corell, 1999).
2.2.3 Road salt
Road salt is commonly used to keep the roads free of ice and snow during winter. Salt lowers the freezing point, making it possible to melt snow and ice down to -18oC. Road salt contains at least 97 % sodium chloride (NaCl). Other constituents include gypsum (CaSO4) and sodium ferrocyanide (Na4Fe(CN)6x3H2O). The sodium ferrocyanide is added to prevent the salt from clumping (Trafikverket, 2015). Road salt has a negative effect on the surrounding environment. For example, increasing chloride concentrations in surface waters may lead to the contamination of drinking water sources, the important processes in plants such as photosynthesis may be inhibited, and soils pore systems may collapse (Blomqvist, 2001).
Also, salt causes a higher fraction of metals to be into a dissolved form, which makes the metals more bioavailable and more difficult to treat (John & Leventhal, 1984).
2.2.4 Total suspended solids
Total suspended solids (TSS) is a measure of the amount of suspended material in water, including both mineral and organic particles. Solids originate from many sources, both anthropogenic and natural. Different sources include the erosion of pervious areas, litter, dust, particles from vehicles and rod wear debris. High concentrations of solids cause increased sediment loadings, siltation and reduce the depth of light penetration within water bodies.
Solids also provide attachment sites, transport and storage for other pollutants (EPA, 1999).
The measure of TSS in stormwater allows for an assessment of sediment transport. It is also an indicator for other pollutants since suspended sediments are an important carrier of, and therefore a surrogate for, a range of organic and inorganic pollutants (Washington State Department of Ecology, 1991)
2.3 Stormwater biofilters
Stormwater management facilities have two main objectives. These are quantity control (e.g.
via infiltration and/or retention) and quality control (i.e. treatment of stormwater pollution).
Below is a description of commonly implemented BMPs (Moreton Bay Waterways and Catchments Partnership, 2006; Water by Design, 2014).
• Vegetated swales utilize overland flow and gentle slopes to convey stormwater instead of, or together with, underground pipe drainage systems. They facilitate settlement and retention and protect waterways by reducing flow velocities
• Buffer strips reduce sediment loads by distributing stormwater as a sheet flow over a vegetated area.
• Retention basins are stormwater detention systems that, apart from managing downstream flow, also offer settling of sediments and associated pollutants.
• Constructed wetlands are artificial shallow vegetated marshes that remove pollutants from stormwater by facilitating a range of processes including sedimentation, fine filtration and plant uptake. In addition, wetlands also provide habitat for wildlife, recreation, ecological values and temporary storage for treated water.
• Infiltration systems (e.g. infiltration trenches, soakaways or basins) capture stormwater and facilitate infiltration into the surrounding soil. This reduces stormwater peak flows and volumes as well as downstream flooding.
• Biofilters (also known as rain gardens or bioretention systems) are shallow depressions that both collect and treat stormwater. Stormwater is treated by filtration through soil media and vegetation combining a variety of physical, chemical and biological processes.
17
It is well established in the literature that urban areas have negative impacts on the quality and quantity of stormwater runoff (Chocat, et al., 2001). Two major goals have been identified related to stormwater management; to retain stormwater quantity (total volume and flood peak) and quality (pollution) as close to pre-development levels as possible (Hatt, et al., 2008). In response, a wide range of stormwater treatment technologies have been developed. There are different infiltration techniques used all around the world that are known for their many advantages. Advantages include improvement of both the quality and quantity of stormwater (Dechesne, et al., 2004). More recently biofiltration systems (also known as rain gardens, biofilters and bioretention systems) have been developed and implemented (Bouwer, 2002).
These systems are receiving increasing interest due to their sizing flexibility, small footprint and aesthetic enrichment. Biofilters have become one of the most frequently used stormwater management tools in urban stormwater treatment (Davis, et al., 2009). There are several published guidelines describing the function and design of biofilters. This study is based on the guidelines from Moreton Bay Waterways and Catchments Partnership (2006).
Biofilters are designed to collect and treat stormwater. As the stormwater percolates through the filter material, filtration, adsorption to substrate and biological processing by soil microbes and plants capture pollutants. The treated stormwater is released into downstream drainage systems. Biofilters also delay stormwater reaching receiving waters by capturing the water in the extended detention zone situated above the filter media surface. Also, the runoff volume may be decreased by evapotranspiration and/or infiltration into surrounding soil if the facility´s design allows (Moreton Bay Waterways and Catchments Partnership, 2006). High flows from storms exceeding design return intervals of 0.5 to 2 years often bypass the system through an overflow pit (Blecken, 2010). Figure 3 shows a model of a biofilter.
Figure 3. Sketch of a typical biofilter (Moreton Bay Waterways and Catchments Partnership, 2006).
The main components of a biofilter are listed below (Water by Design, 2014):
• The filter media supports the vegetation and removes pollutants. It is commonly a mix of loam and sand and is typically 500-1000 mm deep.
• The transition layer is made up of coarse sand and located under the filter media as a bridging layer to prevent finer filter media particles from washing away.
• The underdrainage allows treated stormwater to leave the biofilter.
• An impermeable liner can be included.
• Hydraulic structures consist of an inflow pipe, overflow pit, weir and outlet.
• The role of bunds / embankments is to delay stormwater prior to filtration.
18
• The vegetation (together with the soil) constitutes the biological component of the biofilter. It takes up nutrients and improves the porosity of the soil, supports biological growth and breaks up the filter media surface, which helps prevent surface clogging.
• An inlet pond or sediment forebay provides coarse sediment removal. It detains and stores coarse sediment and disperses energy.
• The maintenance access allows easy and cost effective maintenance.
• The cleanout riser pipe is a pipe connected to each end of the underdrainage pipe. It allows cleaning and inspection of the underdrainage.
Like many other SuD technologies, biofilters have been developed without particular reflection as to how their treatment process is affected by changing season and climate. The treatment efficiency of a biofilter depends on different biological, physical and chemical processes. The extent to which these processes are influenced by variations of the ambient environmental conditions is unknown (Blecken, 2010). Previous studies show high retention of total metals and TSS, with removal generally exceeding 90 % (Davis, et al., 2001), (Davis, et al., 2003), (Lau, et al., 2000), (Muthanna, et al., 2007), (Hatt, et al., 2009), (Blecken, 2010). The results for dissolved metals have been more inconsistent, showing varying removal efficiency (Hatt, et al., 2007), (Muthanna, et al., 2007), (Blecken, 2010).Variable removal of nitrogen has also been shown (Dietz & Clausen, 2006), (Davis, et al., 2006), (Blecken, 2010). Hence, there are still questions regarding the long-term performance and sustainability of biofilters as well as their treatment performance in cold climate conditions.
Although the number of biofilter studies has grown substantially in the last 20 years, many of the processes governing the treatment and hydraulic performance are still not fully understood (Blecken, 2010). Evaluating existing biofilters, their treatment and hydraulic performance would facilitate the development and implementation of reliable and functional biofilters. A biofilter further adapted to varying climate and local environmental conditions would enable implementation of biofilters within the implementation of a sustainable urban stormwater treatment strategy.
19
20 3 Method and materials
3.1 Site description
The studied biofilter is situated on Öringevägen in Tyresö municipality outside of Stockholm (Sweden). The biofilter consists of four smaller biofilters, two on each side of the road. During this study, measurements were performed on two of the four biofilters (identified as A and B; see Figure 4).
Figure 5 shows the biofilter on Öringevägen. The biofilter consists of four sections, two on each side of the road named A and B (right and left in the picture, respectively). A concrete curb surrounds each biofilter at the same height as the pavement. The inlet can be seen located at the front of the biofilter (Figure 5). The picture also shows the macadam wall separating a concrete plate from the rest of the biofilter, creating a micro-sedimentation pond.
During flooding, stormwater can enter overflow pits located in the middle of each filter bed which directly discharges to an outflow pipe connected to a manhole.
Figure 5. Photograph of the Öringevägen (Tyresö) biofilter.
The 26 m2 biofilter represents 0.7 % of the total catchment area (3700 m2). However, the effective catchment area (i.e. that delivering runoff to the monitored biofilter) is approximately 0.19 ha, equivalent to 0,095 ha per biofilter. Approximately 60 % of the catchment area consists of paved road and the remainder is pavement. The area around Öringevägen has Figure 4. Öringevägen (Tyresö) biofilter divided into four smaller biofilters.
21
approximately about 300 households with an estimated traffic intensity of 1000-2000 vehicles/day.
3.2 The structure of Öringevägen (Tyresö) biofilter
The biofilter was built in Tyresö Municipality in 2012. The design involves channelling stormwater from a residential street into the system where it can percolate through a plant- covered filter bed (Larm & Banach, 2012). Once within the biofilter, stormwater may evapotranspirate, percolate through the filter media down to a basal drainage pipe or flow into the overflow pit (in case of high inflow intensities). The overflow pit is connected to a drainage pipe as the surrounding soil is composed of clay. The drainage pipe has a dimension of 110 mm and lies at a 5‰ incline. It lies on a 100-150 mm thick bed and is surrounded by macadam in the fraction size of 16-22 mm. The inlet layer of macadam is in total 350 mm thick with a pore volume of 30 %. A 160 mm outflow pipe connects the overflow pit with a NB1000 manhole in which flow proportional sampling of the outflow is conducted. The manhole is then connected to the existing pipe system (Larm & Banach, 2012).
The concrete curb around the biofilter is formed as an L-support with the same height as the pavement. The biofilter itself is inclined (1:1) such that its highest point is level with the pavement and its lowest point is 0.5 m below the surface soil level (Larm & Banach, 2012).
The filter media on Öringevägen is 500 mm thick, with 5-6 % mold and a porosity of 12 %. The filter media is a proprietary product (Tyresöjord). Particles have an average diameter of 0.2-2 mm meaning that the filter media contains medium to very coarse sand. The filter media is situated on top of a liner made of coconut fibres, which separates the filter media from the macadam. The percolation rate of the filter media is 7 mm/hour. This means that if the biofilter is filled with 300 mm of water it will take 42 hours to drain (Larm & Banach, 2012). The biofilter is constructed with a liner and an underdrainage to avoid infiltration of stormwater into surrounding substrate. A layer of geotextile is placed at the bottom towards the surrounding soil. This layer works as a separating layer between the macadam in the sewer trench and the surrounding soil. A 100 mm thick layer of sand with fraction 0.2-2 mm is placed on top of the EPDM geotextile to even out the ground to prevent the liner being punctured (Larm & Banach, 2012).
De-icing is performed within the Öringevägen catchment using gravel mixed with salt. Seven different plants which naturally occur along the Swedish coastline were planted in the biofilter at a density of 6-9 (bigger plants) or 12-20 (smaller forb plants) per square meter. A concrete plate is placed at the inlet of the biofilter and a 300 mm high wall of macadam (fraction 63-90 mm), separates the concrete plate from the rest of the biofilter. The wall of macadam has an incline of 1:1 towards the biofilter. This separation creates an inlet zone (which works as a micro-sedimentation pond (Larm & Banach, 2012).
3.3 Equipment setup
A 674 ISCO Rain Gauge was installed adjacent to the biofilter to automatically measure and log precipitation at intervals of 0,0254 cm rain. Collection of water samples and flow data involved the installation of a 6712 ISCO Portable Sampler at the inlet (see Figure 6). The 6712 Portable Samplers can sample up to 24 samples. Each portable sampler contains 24 plastic one-litre bottles.
22
Figure 6 Plastic tube equipped with metal intake piece welded on a stainless plate placed at inlet between stainless wall and biofilter curb.
Collection of samples and flow data at the outlet involved the installation of a 200 mm ISCO V-notch together with a 730 ISCO Bubbler Flow Module and a 6712 ISCO Portable Sampler. Both portable samplers were powered by car batteries.
3.4 Sampling
3.4.1 Water sampling
Precipitation proportional stormwater samples have been collected from biofilter A and flow proportional stormwater samples collected from both biofilter A and B.
Sampling programme 1 Tyresö In (automatic sampling unit at the inlet of biofilter A) was initially programmed to start collecting samples when rainfall exceeded ≥1 mm during one hour and the portable sampler located at the combined outlet of biofilters A and B (Tyresö Out) was programmed to collect samples when the water level at the outlet was 0.05 m and the flow was 0.1 litre/s. However, initial analysis of the data indicated that neither Tyresö In nor Tyresö Out sampled complete stormwater events and therefore both Tyresö Out and Tyresö In were reprogrammed after the first six rains. The V-notch connected to Tyresö Out also required adjustment following identification of a range of installation issues.
Sampling programme 2 To address sampling issues identified, Tyresö In was programmed to activate when it rained 1 mm during one hour and to take one sample when it rained 0.3 mm during 15 minutes. Tyresö Out was programmed to activate when the stormwater level was
>0.06 m and when the flow was> 0.1 l/s.
Sampling programme 3 After the eighth rain event Tyresö Out was again reprogrammed to try to enable the capture of complete rainfall events. Samples were therefore collected at a flow interval of 600 litres with a view to capturing the rain event profile.
In total, ten rain events were sampled using three different sampling programmes. The first four rain events were sampled using the plastic bottles. The remaining six rains were sampled using sterile disposable plastic bags placed in plastic containers inside the portable samplers as the original plastic bottles could not be adequately cleaned in the field. After each rain event
23
stormwater from each filled sample bottle was decanted into two new bottles (125 ml and 875 ml), labelled and sent to ALS Laboratory in Danderyd for the analysis of selected metals and nutrients.
3.4.2 Filter media sampling
Filter media from biofilter A was sampled on the 10th of November 2015. In total 18 samples of filter media were collected. Samples were taken from three different locations and at two different depths in the biofilter. At each location and depth three samples were collected. The two sampling depths were 1. the surface of the biofilter (0-5 cm) and 2. the middle of the biofilter (25-30 cm). For more details see Figure 7.
Figure 7. Sampling locations and sampling depths (top picture is an aerial view and the bottom picture a cross-section).
The filter media was sampled using a 40 cm long plastic pipe with a diameter of 3 cm, a 45 cm long wooden stick, a small plastic shovel and plastic bags. The plastic pipe was marked at 5 cm and 25 cm - 30 cm to represent the two sampling depths. Samples were collected by pushing the plastic pipe into the ground of the biofilter media. Filter media trapped inside the plastic pipe was then pushed out with the wooden stick into a plastic bag. The plastic bags were marked according to location and depth of the sample. The numbers 1-3 (see Figure 7) represents the sample location. All filter media samples were placed into a bigger plastic bag and transported to Luleå University of Technology for further analysis of metals and particle size distribution.
3.5 Analysis
3.5.1 Water samples
Samples were analysed for total and dissolved chromium, cadmium, copper, nickel, lead and zinc: For metals, collected samples were filtered through a 0.45 μm filter and then acidified by adding 1 ml Nitric acid (Suprapur) per 100 ml sample. Prepared samples were analysed by ALS laboratory using:
• ICP-SFMS according to SS EN ISO 17294-1, 2 (mod) EPA-method 200.8 (mod),
• ICP-AES according to SS EN ISO 11885 (mod) and EPA-method 200.7 (mod).
24
For the analysis of nutrients (nitrogen and phosphorous), spectrophotometric techniques (CSN EN ISO 6678 and CSN ISO 15681-1) were used for the analysis of phosphorous and IR detection (as per CSN EN 12260) for the analysis of nitrogen.
3.5.2 Filter media analysis
Filter media samples were dried in an oven for 24 hours at 105 °C. Five grams (dry weight) of each filter media sample was then placed in a plastic bag and sent to ALS laboratory in Luleå for metals analysis. Filter media samples were microwave digested using 5 ml nitric acid and 0.5 ml hydrogen peroxide in sealed teflon containers. For analysis of phosphorous and nitrogen 0.1 g dried sample was mixed with 0.4 g lithium metaborate (LiBO2) and then dissolved in nitric acid. Thereafter the treated samples were analysed using the two methods ICP-SFMS and ICP-AES mentioned earlier. To conduct a particle size distribution, samples were dried in an oven at 105 °C, crushed with a mortar and pestle before being placed in a seven-sieve sieving machine for 5 minutes (see Table Error! Reference source not found.1 for sieve sizes). Thereafter t he mass of sediment trapped in each sieve was weighed.
Table 1. Overview of sieve mesh sizes Mesh sizes of sieves (mm)
>2 1.120-2 0.5-1.120 0.25-0.5 0.125-0.25 0.063-0.125
<0.063
3.6 Data Analysis
All data was received from ALS and analysed in MS Excel. Precipitation and flow data logged by the portable samplers onsite was downloaded and analysed using Flowlink. For each rain event, data on stormwater flow depth, amount of precipitation, sampling volume and duration of each rain event is presented in the Appendices together with Flowlink graphs. For each rain event the pollutant concentrations from both Tyresö In (inflowing stormwater) and Tyresö Out (outflowing stormwater) were summed together and an average for each parameter calculated to enable data to be presented as the event mean concentration (EMC).
25
26
4 Results
During the period from July to November 2015, ten rain events were sampled. Table 2 (below) shows the corresponding sampling dates and rain events. Table 2 below shows the rainfall depth associated with each event, their respective sampling dates and identifies when samples were collected at the inlet and/or outlet. Depending on the amount of precipitation that fell during each rain event different numbers of samples were taken.
Table 2. Details of the ten sampled rain events and overview of when samples collected at the inlet and outlet.
Rain event
Rain depth (mm)
Samples collected at Tyresö In
Samples collected at Tyresö Out
1 9,8 16/7/15 16/7/15
2 0,8 18/7/15 No samples taken
3 1,2 21/7/15 No samples taken
4 0,5 No samples taken 30/7/15
5 8,5 2/8/15 2/8/15
6 5,9 5/8/15 No samples taken
7 6,4 15/9/15 15/9/15
8 1,3 No samples taken 16/9/15
9 2,5 7/11/15 7/11/15
10 7,1 8/11/15 8/11/15
4.1 Water quality
Figure 8 – 13 give an overview of the inflow and outflow event mean concentration (EMCs) for selected metals and nutrients during the five rainfall events for which inlet and outlet samples could be collected. Cadmium concentrations were consistently below the limit of detection and therefore are not reported. EMCs show considerable variation – both between replicates (especially for event one), rain events (concentrations in event one typically higher than those determined in other events) and between metals (e.g. chromium generally shows an increase whereas concentrations of nickel generally show a decrease). However, comparison of inflow and outflow EMCs do not suggest a clear pattern for any events or any parameters. Nutrient EMC data show a similar variable pattern of behaviour (Figure 13).
Figure 8. Inlet and outlet event mean concentrations (±SD) for total (left hand figure) and dissolved chromium (right hand figure) for sampled rain events.
27
Figure 9. Inlet and outlet event mean concentrations (±SD) for total (left hand figure) and dissolved (right hand figure) copper for sampled rain events.
Figure 10. Inlet and outlet event mean concentrations (±SD) for total (left hand figure) and dissolved (right hand figure) nickel for sampled rain events.
Figure 11. Inlet and outlet event mean concentrations (±SD) for total (left hand figure) and dissolved (right hand figure) lead for sampled rain events.
28
Figure 12. Inlet and outlet event mean concentrations (±SD) for total (left hand figure) and dissolved (right hand figure) zinc for sampled rain events.
Figure 13. Inlet and outlet event mean concentrations (±SD) for nitrogen (left hand figure) and phosphorous (right hand figure) for sampled rain events.
Table 3 presents the change in concentration between inlet and outlet as a percentage for the 5 events for which both inlet and outlet samples could be collected. A positive value indicates a decrease in concentration between the inlet and outlet whereas a negative value indicates an increase in concentration. Cadmium concentrations were consistently below the limit of detection and therefore are not reported
Table 3. Comparison of metal inlet and outlet EMCs (presented as a %).
Rain Events Parameters
1 5 7 9 10 Mean (±SD)
Cr Total 42 -187 -16 -16 -297 -95 (±142)
Cr Dissolved -1122 -474 -192 -382 -19 -438 (±421)
Cu Total 51 43 3 -12 -106 -4 (±63)
Cu Dissolved 36 48 -3 -38 17 12 (±34)
Ni Total 80 69 23 2 -320 -29 (±166)
Ni Dissolved 80 68 22 11 38 44 (±29)
Pb Total 71 27 43 18 -266 -21 (±138)
Pb Dissolved 74 27 36 -18 26 29 (±33)
Zn Total 69 48 4 -25 -88 2 (±62)
Zn Dissolved 52 31 -5 -26 29 16 (±31)
P 64 84 32 -12 21 38 (±37)
N 43 62 39 -7 -40 19 (±42)
29
Data presented in Table 3 shows considerable variation with all parameters showing both relative increases (negative values) and decreases (positive values) in EMCs on transport through the biofilter. Likewise, the behaviour between all events per parameter shows both increases and increases. In particular events 9 and 10 show an increase in concentration.
Whilst these evaluations are based on concentrations rather than loads, the magnitude of events (all <10mm) generated relatively small inlet volumes. As volumes are relatively low, even small variations in flow (as a result of pollutant release, resuspension, ponding or short- circuiting within the system) can contribute to substantial relative increases in concentration.
The fact that five of the ten rain events sampled did not generated flow at the outlet indicates the challenges encountered during the sampling campaign. For further details see the Appendices.
4.2 Comparison between studies conducted in 2013 and 2015
In 2013, the Öringevägen biofilter was also the subject of a modelling and monitoring campaign, whereby inflow concentrations were predicted using StormTac and outflow measured in the field using flow-proportional sampling. Table 4 shows the results from this study (Larm, 2013).
Table 4. Results from the 2013 Öringevägen biofilter study.
Vol.
(m3)
P* N* Pb** Cu** Zn** Cd** Cr** Ni** Hg** SS* Oil*
Inlet concentrations 1000
vehicles/day
0,1 2,4 3,5 22 38 0,28 7,3 4,2 0,08 65 0,78 2000
vehicles/day
0,1 2,4 4,1 23 46 0,28 7,5 4,4 0,08 67 0,78 Outlet concentrations
Sampling 1 56 0,06 0,51 1,4 31 72 <0,05 2,2 2,5 <0,02 6,9 0,13 Sampling 2 33 0,03 0,22 <0,50 12 52 0,11 1,5 1,8 <0,02 7,7 <0,05 Sampling 3 21 0,06 0,37 1,1 21 54 0,07 1,8 2,3 <0,02 23 0,20 Sampling 4 78 0,04 <0,1 1,1 8,4 42 <0,05 3,5 1,6 <0,02 25 0,10
Key: Vol = volume; * = mg/l; ** = μg/l
Table 5 presents results from the current sampling programme together with the data generated from the 2013 sampling campaign. Data is presented total event mean concentrations (EMCs) averaged for all rain events (i.e. a site mean EMC).
