Metallspridning via dagvatten är ett problem, vilket fo ̈rväntas växa i framtiden. Det finns dock en problematik i att många metoder för dagvattenrening inte är tillräckligt effektiva gällande avskiljning av lösta metalljoner. Detta har medfört ett ökat intresse för reaktiva filtermaterial vilka avskiljer lösta metalljoner. I dagsläget finns en begränsad kunskap om funktionen av reaktiva filter speciellt under nordiska förhållanden med kalla vintrar där dubbdäck är regel och halkbekämpning med vägsalt är vanligt. Detta arbete är genomfört i ett fullskaleförsök under vinterförhållanden med höga salthalter och låga temperaturer i vägdagvattnet. I undersökningen har inverkan av pH och vägsalt på avskiljningen av metaller och igensättning studerats för det reaktiva filtermaterialet Filtralitea® P. Undersökningen har genomförst vid Lilla Essingens dagvattenreningsanläggning som behandlar dagvatten från ett avrinningsomr ̊ade med en yta på 17 600 m2, varav Essingeleden med en årsdygnstrafik ( ̊ADT) på 140 000 fordon, utgör en större del av avrinningsområdet. Undersökningen genomfördes under perioden 28/01-2021 till 16/04-2021. Efter sedimente- ring pumpades vägdagvatten till två parallella filterbrunnar (Filter 1 och Filter 2) laddade med Filtralite P. Uppehållstiden i Filter 1 var fyra timmar medan uppehållstiden i Filter 2 varierade mellan 9 till 24 timmar. Den hydrauliska belastningen på Filter 1 var således avsevärt högre än för Filter 2. Turbiditeten hos det inkommande vattnet till brunnarna var väldigt låg (<46,5 FNU) och igensättning kom inte att påverka drifttiden för filtermaterialet i denna anläggning i någon större utsträckning, då tendenser till igensättning endast påvisades i Filter 1. Kloridkoncentrationen i vägdagvattnet varierade mellan 119 - 4 990 mg L−1 och temperaturen i vattnet som pumpades till de reaktiva filtren varierade mellan 0 – 6 °C. Det reaktiva filtret visade goda förmågor att avskilja partikulärt och löst zink och koppar. Järn- mangan- och kalciumkoncentrationerna avskiljdes i Filter 2, men i Filter 1 ökade ofiltrerade halten järn samt mangan- och kalciumkoncentrationerna. Detta tros bero på att den högre belastningen på Filter 1 orsakade en kraftigare pH-reduktion som i sin tur minskade adsoptionsförmågan och därmed ökade saltets påverkan. En viss avskiljning av krom observerades med undantag fär den filtrerade fraktionen efter Filter 2. Nickelkoncentrationen i utgående vatten var förhöjd i bägge filter (Filter 1: totalhalt max 9,48 μg L−1, löst halt 10,8 μg L−1, Filter 2: totalhalt max 4,08 μg L−1, löst halt 4,13 μg L−1). Koncentrationerna av magnesium och molybden var högre efter båda reaktiva filtren. Metallerna bly, kadmium och kvicksilver förekom endast i låga koncentrationer under försöken. Filter 1 påvisade i regel sämre avskiljning än Filter 2, vilket kan innebära att en lägre hydraulisk belastning resulterar i effektivare rening. Salthalten korrelerade med koncentrationerna av kalcium, järn, mangan, magnesium och zink efter filterbrunnarna och det är därmed möjligt att salthalten kan mobilisera bundna metaller. Ett tröskelvärde för mobilisering av metaller kan vara en salthalt av 2 500 mg L−1. / The growing infrastructure creates more impermeable surfaces, causing an increase in stormwater that brings contaminants from surfaces such as roofs and roads, spreading these in the environment. There are a number of different stormwater treatment methods but many lack the ability to remove metal ions. One promising method is the use of reactive filter materials. However, there is a limited amount of research conducted on how these reactive filter systems perform in natural conditions, especially in nordic environments with cold winters, de-icing salt application and usage of studded tyres. This study aims at in- vestigating how de-icing salt (NaCl), pH, temperature, clogging and water flow affect the treatment efficiency of the reactive filter material Filtralite® P in a real-life application. The catchment area that contributed water to the studied system had a total area of 17,600 m2, including parts of Essingeleden, one of the most trafficked roads in Sweden with an annual average daily traffic load of 140,000 vehicles. The studied system consisted of a stormwater pond fed with water from the catchment area. The pond connected to a measuring well which in turn was connected to a pumping well. From there, water was pumped upwards through two parallel wells (Filter 1 and Filter 2) containing Filtralite P and was then discharged into lake Ma ̈laren. Filter 1 was subjected to a higher hydraulic load compared to Filter 2, resulting i a residence time of four hours in Filter 1 and 9 - 24 hours in Filter 2. Water samples were collected in the measuring well and above the filter media in both reactive filter wells. This allowed for examination of the metal concentrations before and after the reactive filters. Measurements of pH, temperature, conductivity and turbidity were conducted in the measuring well and in both filter wells. In addition, measurements of head loss and water flow were also taken in the filter wells. The samples and measurements of this study were taken between 28/01-2021 and 16/04- 2021. Before that, the reactive filter material had been in use from June to December of 2020. The turbidity in the measuring well varied between 2 and 50 FNU. A decrease in turbidity was observed after the reactive filter media, with an average of 3.3 FNU in Filter 1 and 0.9 FNU in Filter 2. The amount of suspended solids in the water was estimated to not exceed 16 mg L−1 and in total, approximately 1.4 kg of suspended solids was removed in Filter 1 and 0.3 kg in Filter 2. The head loss through Filter 2 was highly correlated with the hydraulic load due to the drastic variations in flow. The head loss in Filter 1 was observed to increase slightly over time, indicating that clogging might start to take place. However, clogging of the filter media did not seem to be an issue for the system, a theory strengthened by the low turbidity and low increase in head loss. The temperature in the measuring well varied between 0 and 6 °C and in the filter wells between 1.1 and 7.4 °C. The low temperature might have negatively affected the efficiency of the reactive filter. An increase in pH was observed after the water passed through the reactive filters, with Filter 1 raising the pH to around 9 and Filter 2 to around 10 in the beginning of the testing period. This increase in pH would diminish over the period in Filter 1 to approximately 8, while Filter 2 maintained a stable pH increase. The conductivity increased over the measuring period and reached a peak in February to then decrease during March, most likely caused by increased and decreased use of de-icing salts. Chloride concentrations in the incoming stormwater varied between 119 - 4 990 mg L−1. The treated water after the reactive filters exhibited a reduced concentration of filtered copper and zink. Meanwhile, the concentrations of calcium, iron, manganese and nickel increased in Filter 1, but decreased in Filter 2. Filtered chromium decreased in Filter 1 but no statistically significant reduction was observed in Filter 2. However, the reduction of chromium might have been more efficient, due to a majority of measurements showing concentrations below the detection limit after the reactive fitlers. There was an increased concentration of filtered molybdenum and magnesium after the reactive filters. The metals cadmium, lead and mercury were merely present in very low concentrations or below the de- tection limit of laboratory instruments. Futher investigations of these metals were therefore not conducted. In general, Filter 2 achieved a higher metal treatment efficiency compared to Filter 1. This is believed to be caused by the higher hydraulic load in Filter 1. This caused a pH reduction in the filter material which in turn led to a reduced adsorption efficiency and possibly enhanced the competition effect between the dissolved metals and the sodium ions from the de-icing salts. A correlation was found between chloride and calcium, manganese and iron in Filter 1. In Filter 2 there was a correlation between chloride and calcium, unfiltered magnesium and zink. The chloride concentrations are related to the de-icing salts and it is believed that the salts could have an impact on the efficiency of metal reduction in Filtralite P. At chloride concentrations of around 2 500 mg L−1 the concentrations of zink, magnesium, iron, copper and calcium seemed to increase in the effluent water, meaning that there might also be a threshold value, which if passed, leads to increased metal mobilization. The higher performance of Filter 2 might also be due to longer residence times. This increa- sed residence time might have allowed for a greater reduction of metals due to an increased contact time. It is important to note that more bed volumes had passed Filter 1 compared to Filter 2, with a total of 321 bed volumes through Filter 1 and 53 bed volumes through Filter 2 during the period of this study. However, if previous bed volumes are also taken into account, a total of 1,157 bed volumes have passed Filter 1 and 559 have passed Filter 2. This could also have caused the reduced efficiency observed in Filter 1 compared to Filter 2.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:kth-298596 |
| Date | January 2021 |
| Creators | Agewall, John, Wallgren, Kim |
| Publisher | KTH, Hållbar utveckling, miljövetenskap och teknik |
| Source Sets | DiVA Archive at Upsalla University |
| Language | Swedish |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | TRITA-ABE-MBT ; 21338 |
Page generated in 0.004 seconds