| dc.description.abstract | Aquifer storage and recovery (ASR) can be an important technique to help meet the increasing freshwater demand in coastal areas. It involves the infiltration and storage of freshwater in a suitable aquifer through a well during times when water is available and its recovery from the same well during times of need. Conventional ASR with a fully penetrating well (FPW-ASR) is generally unviable in brackish aquifers due to the buoyancy force on stored freshwater. However, field pilots suggest that dedicated multiple partially penetrating wells (MPPW-ASR) have the potential to increase the recovery efficiency (RE) of ASR in brackish aquifers. The present study focused on validating and quantifying this improved recovery with use of a 2D lab-scale sandbox model, complemented by numerical SEAWAT simulations. A representative numerical 3D field-scale model with a normal fully penetrating well was able to validate the reduced RE of FPW-ASR with a lower dimensionless D-parameter, i.e. with a larger relative influence of buoyancy. The RE also reduced with a higher ratio of injection radius to aquifer height (r/H), i.e. with a longer duration of infiltration. The latter observation contradicts previous studies, possibly due to numerical dispersion, and requires further research. Nevertheless, simulating ASR scenarios with different aquifer properties and operational parameters can result in the same RE, provided that D, r/H, and the relative durations of the ASR phases are constant for a given grid discretization. This enabled downscaling of numerical ASR scenarios from field-settings to lab-settings. The improved RE of freshwater using MPPW-ASR instead of FPW-ASR in brackish aquifers was confirmed by the numerical 3D-field scale model. However, the RE of MPPW-ASR was not necessarily optimised because of the simplified operation in this study. When MPPW-ASR is optimised for a specific scenario, the RE can be further ameliorated. However, the potential relative improvement decreases with a less significant buoyancy effect (a lower r/H and a higher D), because the RE of conventional ASR increases. The 2D sandbox was able to validate this performance. However, the results of a 3D-setting with a linear well were not transferrable to those of a 2D-setting with a planar well due to the difference in nature of both settings. Moreover, the experiments consistently underestimated the beneficial effect of MPPW-ASR that was observed in the numerical field-scale model with a vertical planar well. This probably resulted from the combined effect of measurement inaccuracies, heterogeneities in the aquifer packing, increased mixing due to the experimental well configuration, and (numerical) dispersion. The sandbox should thus be considered mainly as a validation and visualisation tool rather than a quantification tool. | |
