Water quality deterioration is a common phenomenon that may limit the recovery of injected water during aquifer storage and recovery (ASR). Quality deterioration is often caused by the oxidation of reduced aquifer components by oxygenated source water, the subsequent pH decline, and induced dissolution of carbonate minerals. We use a previously calibrated reactive transport model (PHREEQC) to optimize ASR depending on source water quality and kind of pretreatment. We give quantitative projections on the performance increase over successive cycles with respect to specific water quality indicators. We simulate the response of a representative, deeply anoxic aquifer upon injection of three different commonly applied oxygenated water types: pre-treated drinking water, desalinated seawater, and urban storm water. The model is coupled to a Python script that automatically stops recovery and starts the next injection phase when certain specified concentration thresholds are exceeded. This setup enables realistic simulations to gradually create a buffer zone around the ASR well that allows 100% recovery at a specific stage of aquifer development. Each source water type was associated with different issues causing the deterioration of the abstracted water quality with respect to Fe(II), Mn(II), and As. The injection of pre-treated drinking water caused Mn(II) exceedances that disappeared after a number of cycles, provided that the recovery would halt as soon as the Mn(II) exceedances would occur. The injection of desalinated water caused persisting Fe(II) exceedances, which substantially slowed the creation of an effective buffer zone; whereas, the injection of urban storm water caused similar issues with respect to arsenic. For both cases, it was shown that enriching the source water with O2 and/or NaOH had major positive effects by accelerating the creation of an efficient buffer zone. Finally, we simulated a long-term operational rest of the ASR plant to evaluate water quality effects during potential migration of the stored water due to lateral groundwater flow, as dependent on source water composition and pretreatment method. Exceedances of drinking water guidelines occurred long before the arrival of the native water. Fe(II) and Mn(II) exceedances, after having used desalinated and drinking water, respectively, were observed after a bubble migration of 9% and 40%, respectively, and were associated with the slightly acidic pH conditions promoting the dissolution of Mn-carbonate and preventing an efficient sorptive removal. The arsenic exceedances, after using urban storm water, were associated with the arsenic wave deriving from the pyrite oxidation and reaching the ASR well after 34% of bubble migration. Enrichment of the source water with O2 and/or NaOH was also helpful in protecting the water quality around the ASR well for a longer time during a bubble migration scenario. The Fe(II) breakthrough occurred after 59% of desalinated bubble migration (instead of 9%) whereas As broke through after 70% of urban storm bubble migration (instead of 34%). This study illustrates that reactive transport modeling with a calibrated model is a useful tool to a-priori test the potential effectiveness of various operational options in ASR application on improving recovered water quality and the recovery efficiency.@en

Water quality deterioration is a common occurrence that may limit the recovery of injected water during aquifer storage and recovery (ASR) operations. This limitation is often induced by the oxidation of the reduced aquifer components by the oxygenated injection water. This study explores the potential of aquifer pre-oxidation using permanganate to improve the quality and volume of the recovered water during ASR. An experimental ASR column setup was developed to simulate the oxygenated water injection and recovery cycles. Undisturbed sediments from an anoxic brackish aquifer at a pilot ASR site were used. A series of 4 conventional ASR cycles injecting oxygenated tap (drinking) water was initially performed. These experimental trials showed a persistent Mn(II) production due to the dissolution of a Mn-containing carbonate that was triggered by pyrite oxidation reactions, as shown by the observed sulfate production. The rise in the Mn(II) concentrations above the drinking water standards would limit the recovery to 15–30% of the injected water without treatment of the recovered water. To a lesser extent, arsenic production resulting from the oxidative dissolution of pyrite posed a water quality threat to the ASR operation. Consequently, a second series of experiments was performed with an oxidation cycle using a dilute (5%) potassium permanganate (KMnO4) solution, aimed at deactivating the reactive phases responsible for the acidity-triggered Mn(II) production. This pre-treatment cycle improved the net recovery ratio to 84% during a conventional ASR cycle using oxygenated tap water. The extent of pyrite oxidation was decreased by 63% after the permanganate treatment. The increased competition for oxygen by the adsorbed Fe(II) and Mn(II) on the newly precipitated Mn-oxides combined with the pyrite “deactivation” by removal of the most reactive iron-sulfide crystals during the permanganate flush were primarily responsible for the observed decrease in the pyrite oxidation. The stability of the Mn-oxide precipitates was tested by flushing the columns with native groundwater before performing a final ASR cycle, simulating an increase in the reducing conditions during static periods in the ASR system. The Mn-oxide reduction by ferrous iron in the native groundwater released substantial amounts of Mn(II), which adversely affected the ASR operation in the subsequent cycle. In these cases, repeating the permanganate treatment should be considered.@en

This reactive transport modeling study presents a follow up to the mass balance-based identification and quantification of the main hydrogeochemical processes that occurred during an aquifer storage and recovery (ASR) trial in an anoxic sandy aquifer (Herten, the Netherlands). Kinetic rate expressions were used to simulate oxidation of pyrite, soil organic matter (SOM), and ferrous iron, and dissolution of calcite and Mn-siderite. Cation exchange, precipitation of Fe- and Mn-(hydr)oxides, and surface complexation were treated as equilibrium processes. The PHREEQC model was automatically calibrated with PEST to observations from the first ASR cycle, and was then allowed to run for all 14 cycles to evaluate its long term performance. A sensitivity analysis was conducted to find the most controlling model parameters. Pyrite was ranked as the most important reductant, followed by SOM, whereas Fe(II) was least important. Moreover, the pH and oxygen gradients were found to enhance the rate of pyrite over SOM oxidation with distance away from the ASR well. The increasing sorption capacity of precipitating Fe-hydroxides was reflected by the decreasing Fe(II) concentrations with subsequent cycles whereas Mn(II) showed a tendency to mobilize during recovery and remain above standards. Oxidation and dissolution rates were found to depend on travel time and injection rate as well as on the presence or absence of flow. Oxygen enrichment of the injection water increased oxidation rates and therefore accelerated the aquifer’s leaching from its reactive species. We specifically focused on impeding the release of Mn(II) to the groundwater, a process that acted as a restraining factor for the feasibility of ASR application at this site. The undesirable side-effects of oxygen enrichment as well as the Mn(II) issues were found to be partly suppressed by enriching the source water with pH buffers according to scenario simulations.@en

The hydrogeochemical processes that took place during an aquifer storage and recovery (ASR) trial in a confined anoxic sandy aquifer (Herten, the Netherlands) were identified and quantified, using observation wells at 0.1, 8 and 25 m distance from the ASR well. Oxic drinking water was injected in 14 ASR cycles in the period 2000–2009. The main reactions consisted of the oxidation of pyrite, sedimentary organic matter, and (adsorbed) Fe(II) and Mn(II) in all aquifer layers (A–D), whereas the dissolution of carbonates (Mg-calcite and Mn-siderite) occurred mainly in aquifer layer D. Extinction of the mobilization of SO4, Fe(II), Mn(II), As, Co, Ni, Ca and total inorganic C pointed at pyrite and calcite leaching in layer A, whereas reactions with Mn-siderite in layer D did not show a significant extinction over time. Iron(II) and Mn(II) removal during recovery was demonstrated by particle tracking and pointed at sorption to neoformed ferrihydrite. Part of the oxidants was removed by neoformed organic material in the ASR proximal zone (0 – ca. 5 m) where micro-organisms grow during injection and die away when storage exceeds about 1 month. Anoxic conditions during storage led to increased concentrations for a.o. Fe(II), Mn(II) and NH4 as noted for the first 50–200 m3 of abstracted water during the recovery phase. With a mass balance approach the water–sediment reactions and leaching rate of the reactive solid phases were quantified. Leaching of pyrite and calcite reached completion at up to 8 m distance in layer A, but not in layer D. The mass balance approach moreover showed that Mn-siderite in layer D was probably responsible for the Mn(II) exceedances of the drinking water standard (0.9 μmol/L) in the recovered water. Leaching of the Mn-siderite up to 8 m from the ASR well would take 1600 more pore volumes of drinking water injection (on top of the realized 460).@en