This study addresses the critical aspects of injection well clogging as part of an Aquifer Storage, Transfer and Recovery (ASTR) pilot infiltrating tile drainage water (TDW) from agricultural field into an anoxic brackish sandy aquifer for later re-use as irrigation water for flower bulbs. Two ASTR injection periods suffering from clogging were intensely monitored during several weeks in 2019 to 2020. The wells were rehabilitated by methods of backflushing and mechanical cleaning (high-pressure jetting) and sampled to obtain information on the kind of clogging material. The recharge water is nutrient rich (PO4: 2-18 mgL-1, NO3: 6-50 mgL-1, NH4: 0.1-0.6 mgL-1, DOC: 25-32 mgL-1) and attained significantly turbid conditions (usually between 5-20 NTU; up to 160 NTU) caused by the removal of resuspended biochemical material in the tile drainage network by extreme precipitation and extensive drain discharge events. Consequentially, the 40 µm spin Klin-disc filters as pre-treatment step experienced reduced functionality due to clogging when subjected to high turbid loads. Microbial processes within the ASTR piping system additionally caused fluctuations in turbidity by periodic on/off operation, indicating a high potential for (bio)physical and biological clogging in the injection wells. Microscopic and (hydro)geochemical analysis of injectant and backflushed suspended solids demonstrated a significant contribution of injected Fe-hydroxyapatite flocs and biochemical material incorporating siliceous clay and silt particles below 40 µm, while pyrite and possibly calcite precipitants were flushed from the aquifer matrix when removing suspended matter from the wells. The injection of suspended material caused clogging predominantly by physical mechanisms. During standstill, the hydrochemical environment in the well indicated substantial microbial activity by reduced redox conditions in the well and the mobilization of Fe(II), Mn(II), P, and Ca, indicating the vulnerability for biological clogging. Improving the feasibility of the ASTR system requires the high clogging potential of TDW to be reduced. It is recommended to adopt a settling tank, rapid sand filtration followed by slow sand filtration as pre-treatment steps to reduce the physical and biological clogging potential. Various other recommendations to reduce the clogging risks are discussed further. Regarding the well rehabilitation, methods of compressed-air ‘jutteren’ are recommended to recovery the well performance by effectively removing residual clogging material from the gravelpack and borehole wall.

Pressure on fresh water resources has led to water scarcity and increasing demand for alternative water resources, such as rainwater. Nevertheless, heavy metal contamination is a limiting factor for re-use of urban stormwater. This research focusses on the adsorption of heavy metal contaminants by biofiltration systems, specifically on the Bluebloqs Biofilter, to fulfil the water quality requirements for infiltration. The effect of various physical and chemical conditions on HM adsorption were assessed.A field experiment was built in the summer of 2019 and operated during 5 months. A range of low to high heavy metal concentrations was synthetically dosed to functional prototype of the Bluebloqs Biofilter, with the aim to characterise both the removal efficiency as well as geochemical processes in the filter media. Results showed that due to saturation of the filter media, Cd and Zn concentrations in the effluent rose above the Dutch infiltration standards after 74 bed volumes. Ni, Cd and Zn were most susceptible for bed saturation as average concentrations over biofilter depth increased by operation. Highest removal of all heavy metals happened in the top 5 cm of the filter bed. A surface complexation sorption model was developed in PHREEQC to evaluate adsorption of Cd, Cu, Ni and Zn. The model was established by calibration on batch adsorption experiment data of the same heavy metals. With this modelling tool, various physical and chemical conditions in stormwater could be easily simulated and extrapolated, as well as characteristics of filter media that could be modified. Surface complexation modelling showed a reasonable fit compared to the Freundlich isotherm.The order of adsorption to the filter media was Cu>Cd>Zn>Ni. Batch simulations showed that by raising solution pH, adsorption of each HM was increased for both quartz sand and iron oxide coated sand. For pH 7.0, the partition coefficient for iron oxide coated sand was 1.34, 16.8, 19.9 and 23.9 for Ni, Cd, Cu and Zn respectively. For pH 8.0, the partition coefficient rose to 12.9, 2.15*103, 282 and 698 in the same order. In the case of quartz sand and pH 7.0, partition coefficient was found 4.7 *10-2, 1.18, 1.41 and 14.8 for Ni, Zn, Cd and Cu respectively. When increased to pH 8.0, partition coefficients rose to 0.7, 34.0, 228 and 201 in the same order.To a lower extent, increase of ionic strength of the solution had a negative effect on the adsorption of Cu and Cd. By increasing the Zn and Cu proportion by a factor 5.0 in the influent stormwater, adsorption of Ni and Cd was negatively affected, but only for unrealistic high concentrations.One dimensional transport simulations were performed for the average stormwater compositions found at the field experiment. Ni showed first breakthrough in the filter due to lower affinity, followed by Zn, Cd and Cu. Increase of the pH from 7.0 to 8.0 delayed breakthrough by a factor 9.2, 15.8, 25.9 and 135.1 for Cu, Ni, Zn and Cd respectively, despite the type of media used in simulations. By changing the 20% of the top layer of the total filter length from quartz sand to iron oxide coated sand in the model, the adsorption capacity increased and delayed the breakthrough of heavy metals by a factor ±2.36 at all pH conditions.Comparison of the established sorption model with field experiment data showed underestimation of Ni adsorption, while Cd and Cu were overestimated. Zn adsorption showed the most similarity between model results and field data. The model cannot provide an accurate prediction on HM adsorption, but it can be used to compare in between HMs.For the first mature system of the Bluebloqs biofilter, operational in Spangen, Rotterdam since the summer of 2018, the first exceedance of the Zn for the infiltration standards was predicted after 410 bed volumes, due to saturation of filter media. This resulted in elevated Zn concentration in the effluent. Further improvement of the sorption model could be established by additional batch adsorption experiments to obtain more trustworthy results.