Artificial recharge of aquifers can be performed for various purposes and under varying hydrogeological conditions. We present an overview of deep-well recharge applications which have taken place in the Netherlands over the last two decades. We present the purpose of each application, the issues which had to be resolved, the preventive measures which were taken to improve performance and the lessons learned from each experience. Examples are given of applications which aimed at the storage of water for drinking and other purposes such as irrigation, achieving environmental goals and disposal of wastewater. Applications aiming at drinking water production usually faced issues related to the quality of the abstracted water not meeting drinking water standards with respect to various elements, such as iron, manganese and arsenic. Storage of water in brackish aquifers was complicated by buoyancy effects making part of the recharged water irrecoverable. Recharge of water with the purpose of recovering declined groundwater tables and fighting seawater intrusion was hindered by clogging of the injection well while the disposal of wastewater was limited to aquifers of lower groundwater quality.
Artificial recharge of aquifer systems may be performed for various reasons. These include, among others, the subsurface storage of water for drinking and agricultural purposes, the natural subsurface water treatment, the achievement of environmental goals (e.g. recovery of declining water tables and the prevention of saltwater intrusion), and the disposal of wastewater. The main financial advantages of subsurface water storage compared to surface water storage consist of (1) increased security of the water supply, (2) little land occupation, (3) reductions in costs of water storage and evaporation losses, and (4) during water purification, storing the treated water in the subsurface can decrease the peak factor allowing the facility to have a smaller capacity (Pyne 2005).
In the Netherlands, climate variability has increased the need for water storage as a result of (1) stronger fluctuations in river discharge and groundwater tables and (2) enhanced intrusion of the North Sea along the Dutch coast. Another driver for storing water underground is the increasing demand for drinking water in areas where the government has forced the water supply companies to restrict their groundwater pumping. This is usually enforced in order to prevent the drawdown of groundwater tables, to restore wetlands in selected areas and to prevent seawater intrusion. Moreover, recent benchmarks of drinking water prices are pushing the Dutch water supply companies to lower their costs. Aquifer storage applications, if operating without quality deterioration, are thus becoming an interesting option also in the Netherlands.
Testing of deep infiltration systems in the Netherlands started in the 1930s without much success due to severe clogging of the infiltration wells. Later in the 1970s, a number of water supply companies (PWN, Waternet, Dunea) started integrating deep-well recharge systems as production units with their existing open infiltration systems. Decisive success factors were the prevention of clogging by extra purification of the recharged water and the treatment of clogging by frequent and intensive back-pumping of the wells. Nonetheless, the ideal situation of recovering stored pre-treated drinking water without the need for post-treatment is rarely achieved.
DRINKING WATER PRODUCTION
Since the early 1970s, numerous deep-well injection and aquifer storage experiments have been carried out in the Netherlands to acquire insight on (1) the feasibility of recharging confined aquifers with surface water through wells and (2) the water quality changes upon interaction of the injected water with the native groundwater and aquifer sediments (Stuyfzand 1998). The intentional use of natural attenuation processes to improve water quality (Maliva & Missimer 2010) has been deliberately studied to improve understanding of them and fully harness their potential in combination with the infiltration or storage of water. The environmental profits of these natural purification methods are many, the most important being a reduced application of undesired chemicals like coagulants, active carbon, ozone, and chlorine.
The target aquifers are usually composed of Pleistocene or Miocene sands with a grain size of 300–500 μm and a permeability of 30–50 m/d. In most cases the native groundwater is fresh, anoxic, calcite saturated and with relatively high concentrations of Fe(II), Mn(II) and ammonium. Most formations contain low amounts of very reactive iron sulfides (pyrite). Calcium carbonate is typically present in marine deposits and Pleistocene sediments while small amounts of iron carbonates that contain manganese and/or magnesium (e.g. siderite, ankerite, rhodochrosite) are often present.
Hydrogeochemical processes affecting water quality
Upon injection of aerobic water, the water quality is mainly altered by oxygen (and nitrate) consumption by sedimentary electron donors like pyrite, sedimentary organic matter, and exchangeable Fe(II), NH4+, and Mn(II). The induced acidity is partly buffered by HCO3- in the groundwater and, if present, by the dissolution of carbonate minerals. Carbonate dissolution further affects the water quality due to the possible release of Fe(II) and Mn(II) in the groundwater. Heavy metal release may also result from cation exchange (Ca2+ in the injected water displacing Fe(II) and Mn(II) from the exchanger). If oxygen is present then the released Fe(II) is further oxidized (homogeneous oxidation) and precipitates in the form of Fe-hydroxide (ferrihydrite). If however the oxygen front is lagging behind, as it progressively happens further from the injection well, then Fe(II) remains dissolved and may contaminate the water that is abstracted from a well downstream of the injection well (i.e. ASTR schemes). The kinetics of the homogeneous Mn(II) oxidation are very slow (Diem & Stumm 1984) resulting in dissolved Mn(II) contamination being more persistent, even in the presence of oxygen.
The oxidation of pyrite may also lead to some mobilization of trace elements, such as As, Co and Ni. As especially is more mobile due to its reduced state (arsenite), which prevents it from sorbing on the existing and newly formed Fe-hydroxides. With continuous injection, the advancing oxygen (and nitrate) fronts convert the dissolving As into its oxidized state (arsenate), which is much less mobile due to preferential sorption on Fe-hydroxides.
The evolution of abstracted water quality allows the identification of a sequence of distinct phases, each of them spanning over a number of pore flushes. The gradual leaching of sedimentary electron donors is testified by the gradually decreasing SO4 production and electron acceptor (NO3- and O2) consumption. The removal of reactive phases is also reflected by the decreasing Ca and HCO3- production, indicating a decline in Ca-carbonate dissolution, a process that is normally triggered as a response to oxidation reactions and the induced acidity. As observed in Figure 3, the complete leaching of calcite requires a much longer flushing time as, upon complete leaching of electron donors, calcite dissolution is only triggered by the negative calcite saturation index of the injected water.
Measures to prevent contamination
Field observations on ASR sites and reactive transport modelling have suggested an increasing sorption capacity with subsequent injection–recovery cycles due to a gradual build-up of Fe-hydroxide precipitates. Moreover, a buffer zone can be implemented around the ASR well by halting recovery and restarting injection as soon as drinking water standards are exceeded (Antoniou et al. 2015). This approach shifts the reactive zone further away from the ASR well and can substantially increase the recovery efficiency (i.e. ratio between recovered and injected volume) of the ASR plant. However, if Mn-containing carbonates are present then aquifer pre-treatment or source water pre-treatment may be required to prevent the dissolution of these carbonates during recovery phases and the early exceedance of drinking water guidelines with respect to Fe(II) or Mn(II).
The use of pH-buffering agents such as NaOH and Na2CO3 can increase the alkalinity and pH of the aquifer around the ASR well and prevent the dissolution of carbonate minerals as a response to under-saturated source waters or acidifying water–aquifer interactions. Additionally, increased pH will enhance the sorption of Fe(II) and especially Mn(II) on the Fe-hydroxides formed during injection. The effects of dosing a pH-buffer have been experimentally tested with positive results in real-scale ASR pilots in Virginia and South Carolina which suffered from Fe(II) and Mn(II) concentration exceedances in the recovered water (Ibison et al. 1995; Pyne et al. 2013). Using pH-buffers to increase the aquifer's pH should be avoided if As concentrations are high as adsorbed As (as arsenate) tends to desorb with increasing pH.
Pre-treating the aquifer material was tested by means of column experiments where ASR was realistically simulated in an anoxic setting using real aquifer sediments (Antoniou et al. 2014). It was concluded that treating the aquifer sediments with permanganate can prevent the mobilization of Fe(II) and Mn(II) and greatly improve the recovery efficiency. This is the combined result of (1) the substantial depletion of sedimentary electron donors due to increased oxidation, (2) the extended precipitation of Mn-oxides (by-product of the oxidation reactions) with a high sorption capacity and (3) the increased pH conditions due to proton consumption. Care should be taken so as to prevent the reduction of the Mn-oxides by the inflowing native groundwater during an extended recovery phase or during intentional or circumstantial downtime of the plant, as this may lead to significantly high concentrations of dissolved Mn(II) in the groundwater.
It is expected that ASTR schemes suffering from heavy metal contamination would also benefit from application of pH-buffers although such tests have not been performed to our knowledge. Neutralization of the aquifer material lying between injection and abstraction wells is also beneficial; it should, however, not be performed using permanganate as this would cause severe increases in Mn(II) concentrations in the abstracted water due to the eventual reduction of the newly formed Mn-oxides. A slower neutralization process by injecting water enriched with O2 would be more appropriate.
DEEP-WELL RECHARGE FOR OTHER PURPOSES
Irrigation water production
Subsurface storage of water can also be performed for irrigation water supply. In this case, concentration thresholds of total dissolved solids are less strict than the ones applied for drinking water. This allows brackish or saline aquifers also to be considered for storage and later recovery during high irrigation water demand. A number of recent field experiments in the Netherlands proved that aquifer storage of freshwater in brackish or saline aquifers can be an efficient technique to bridge freshwater shortages in coastal areas (Figure 1). The main challenge in such a setting is dealing with the buoyancy effects which may cause salinization at the bottom of the ASR well during recovery, making part of the fresh water irrecoverable.
A recent study (Zuurbier et al. 2014) was performed in an area dominated by greenhouse horticulture where a high irrigation water demand was until recently covered by storage of rainwater in basins/tanks, use of surface water, and desalination of brackish groundwater (Paalman et al. 2012). The experiment aimed at reducing freshwater losses by applying deep injection and shallow recovery using multiple, individually controlled well screens. This configuration slowed down salinity build-up in the recovery well screens and the recovery efficiency was substantially increased compared to a conventional fully penetrating well.
Achieving environmental goals
Deep-well injection has also been applied in the Netherlands in order to achieve specific environmental goals. In these cases, specific aquifers are usually recharged in order to recover declined water tables to older (higher) levels or to fight salinization due to seawater intrusion.
The sand filter was cleaned once a year to limit the development of hydraulic resistance due to the continuous infiltration of water with moderate quality. Hydraulic resistance developed also in the injection wells mainly due to the accumulation of biomass and iron flocs in the gravel pack. This resistance disappeared with automatic backwashing, switching on as soon as the water level inside the well exceeded a certain threshold. The removal of clogging which developed on the borehole wall (chemical clogging) required occasional well-surging using compressed air. Currently the extra infiltration capacity in the dune aquifers is not needed and the well infiltration system has stopped.
The production of fresh drinking water from brackish groundwater by reverse osmosis (BWRO) is becoming more attractive, even in temperate climates. The drivers in this case may consist of environmental problems like the pollution and salinization of aquifers, drawdown of water tables in phreatic aquifers, effects of climate change like reduced base flows that render surface waters less fit for drinking water production, and increasing costs to produce drinking water from heavily polluted, fresh groundwater.
Two pilots were initiated in 2009 to test the BWRO concept in the Netherlands (Noardburgum and Zevenbergen). The simultaneous abstraction of upper fresh and lower brackish groundwater led to a lowering of the fresh–brackish water interface confirming that instead of lowering production, brackish groundwater should be pumped and used (Raat et al. 2012). At both locations, concentrate injection was technically feasible, as long as the RO recovery levels were not higher than 50% (Zevenbergen) or 70% (Noardburgum). At higher levels, clogging of the injection well due to mineral precipitation could become an issue.
The first step when it comes to replicating these technologies is to assess and map their spatial feasibility for the identification of suitable locations. Predicted spatial suitability was assessed in the Westland-Oostland region, the Netherlands, through comparison with the measured performance of existing sites in the area (Zuurbier et al. 2013). A complete suitability assessment may consist of the integrated analysis of the geomorphological, hydrogeological, socio-economic, and legal conditions pertaining in the region of interest. Products such as suitability maps and information management systems can serve as guidance for interested stakeholders and decision makers in assessing baseline conditions and choosing the solution which best fits to their needs. Information management systems can also help achieve broader replicability and up-scaling of novel solutions as they allow for an interactive and easy-to-follow description of background methodologies.
Although suitability maps serve as a valuable implementation basis, additional real-scale validation should be carried out to confirm the suitability of the proposed locations. As presented in this paper, the success of applications aiming at drinking water production very much depends on the geochemical processes taking place upon aquifer recharge. The presence and reactivity of mineral phases cannot be adequately assessed a priori without piloting or at least laboratory-testing of the aquifer sediments. Fine-scale hydraulics and hydrogeology may also need to be captured to fully assess feasibility at specific locations. It is therefore recommended to always validate maps further via modelling, laboratory-testing and piloting.
The Dutch experiences showcased in this overview highlight the value of deep-well recharge interventions into both fresh and saline aquifers. Depending on the scope, there are different issues to deal with and different preventive measures to consider. Applications aiming at drinking water supply must pay special attention to hydrogeochemical interactions between recharged water, native groundwater and aquifer sediments. Applications in brackish or saline aquifers should consider feasibility with regards to buoyancy issues and mixing effects. All applications, regardless of purpose and groundwater quality, should consider the possible lateral movement of the injected water caused by the native hydraulic gradient. If the recharged water is to be recovered at a later stage, which is usually the case in aquifer storage applications, this lateral drift may complicate abstraction of the stored water and substantially reduce the recovery efficiency. Spatial suitability maps and information management systems can facilitate the identification of suitable locations and intervention types depending on specific needs.
The authors would like to thank two anonymous reviewers for their comments and suggestions which improved the quality of this article.