Filter beds for on-site wastewater treatment Towards more reliable estimations of phosphorus retention

Sammanfattning: Phosphorus (P) is an important plant nutrient and essential for life. However, the phosphate rock used for fertilizer production is a non-renewable resource and its production is expected to peak. At the same time, the discharge of P into natural waters is causing eutrophication, a severe problem in areas such as the Baltic Sea. Onsitewastewater treatment facilities in Sweden contribute substantially to this discharge because of their inadequate retention of P. Filter beds are a potentially useful technique to capture P in on-site facilities. However, many variables need to be considered when the P retention of potential filter materials is estimated in laboratory tests prior to designing full-scale filters. The overall aim of this thesis was therefore to increase the reliability of forecasting P retention in full-scale filters by increasing the understanding of P retention in P filters under varying conditions and by identifying measures that could lead to more reliable methods of testing filter materialsat laboratory scale. The effects of influent type, influent P concentration, loading rate and ambient temperature on the filter materials Filtra P, Filtralite P, Top16 and Polonite were investigated in filter column experiments using 22 factorial designs. Furthermore, the P binding mechanism was studied using mineral phase investigations and by determining the reaction time of the P in the filter. In addition, filter performance was estimated by means of hydro-geochemical transport modelling. The investigated factors significantly (Į = 0.05) affected the retention of P in the filter materials showing that it is important to consider those factors when designing laboratory filter experiments and full-scale filters. Using secondary wastewater as an influent instead of P solution decreased the P binding capacity of Filtralite P, probably due to organic compounds contained in the wastewater. Increasing influent P concentration decreased the number of bed volumes treated before breakthrough in Filtra P by 82%. The loading rate was shown to be an important design parameter. Increasing the loading rate, something commonly done in the laboratory to accelerate the testing, significantly increased the amount of washed-out particulate P in Filtra P and Filtralite P. However, the residence time was also shown to be important; it should be maximised in filter tests as far as practical constraints allow. Increasing the temperature from 4.3 to 16.5°C increased the P binding capacity in both Top16 and Polonite which was attributed to an enhanced precipitation of calcium phosphates. This indicates that results obtained from experimental filters at room temperature might overestimate filter performance in the field where the temperature can be lower. In addition, full-scale filters might function better in warm rather than cool climates. The results further showed that it is crucial to measure both the concentration of dissolved P and particulate P in the filter effluent, as P-containing particles were observed to escape fromthe experimental filters in this study. Hydroxylapatite was detected in the outflow hose of the Filtra P columns indicating that this mineral phase may form in the filters under certain conditions. The geochemical models, however, indicated that the only precipitated calcium phosphate compound was amorphous tricalcium phosphate.Two hydro-geochemical transport models were developed that satisfactorily described the experimentally derived P breakthrough curves and effluent pH of the filter columns with Filtralite P. The simulations suggested that calcium oxide, calcite and the calcium-silicate phase wollastonite supplied the Ca2+ and OH- ions required for the precipitation of phosphate.