Monitoring of the FRESHMAN pilot with AH-DTS and ERT

Introduction and general objectives of the project The FRESHMAN research project was initiated in 2020 to investigate the feasibility of using brackish groundwater as a sustainable source for drinking water production for Dunea. To this end, a field pilot near Scheveningen was carried out to study several aspects of brackish groundwater extraction. From early January 2022 to halfway March 2025, the following four main phases of the pilot took place: - Phase 1: Extraction of brackish groundwater. This extraction for about 20 months led amongst others to a downward expansion of fresh groundwater (freshening). - Phase 2: Extraction of deep fresh groundwater. For about 3 months, deep fresh groundwater was extracted to simulate a calamity in which no river water is infiltrated, whilst the production of drinking water is continued. This resulted in saltwater upconing. An idle period of a few months preceded Phase 2. - Phase 3: Extraction of brackish groundwater. Similar as phase 1, but for a shorter period of time (about half a year), in order to restore the freshwater well screens (downconing). - Phase 4: Simultaneous extraction of fresh groundwater and brackish water extraction. This phase took place for about 3 months, to keep the interface between fresh and brackish groundwater stable (Freshkeeper mode). In addition to the conventional monitoring techniques applied within FRESHMAN, an additional TKI project ‘COASTAR Brakwaterwinning; meer met meten’ was organized. In this TKI project, two monitoring techniques were proposed and implemented in the pilot study; Active Heating Distributed Temperature Sensing (AH-DTS) and cross-hole Electrical Resistivity Tomography (ERT). AH-DTS relies on fiber-optically measuring the temperature response during and after generating a heat pulse, which, in the current project, was mostly used to infer groundwater flow rates. The heat pulse is generated electrically by a resistance wire, which is integrated with the fiber optic cable into a single cable. ERT is based on earth resistivity measurements, which are influenced by lithology and groundwater quality (salinity). Time-lapse ERT measurements can be used to monitor salinity dynamics, as it can generally be assumed that lithological changes don’t change significantly over time. The main objectives of AH-DTS and ERT were to gain better insight into 1) three-dimensional spatiotemporal variations in salinity and 2) options for improving operational management of brackish groundwater extraction. This report describes the methods and findings of this TKI project. Methodology ERT cables were installed at four different locations (including two extra boreholes) and AH-DTS cables at four locations along the depth of interest in the subsurface. All downhole AH-DTS and ERT cables were connected through surface cables to in-house and commercial monitoring hardware, situated in a central control shed. Using this monitoring network, over 10000 cross-hole ERT measurements were collected during the pilot with very few interruptions every 8 hours, using 48 electrodes per measurement location at an electrode space of 1.67 m. For the AH-DTS, data were collected without significant hiccups during the first two years, but were significantly discontinued due to electrical issues during 2024, except for the last few months. For selected periods, the AH-DTS data were used to infer groundwater flow velocity towards the well screens based on a relation between peak temperature during a heat pulse and corresponding groundwater flow rate. Results The AH-DTS data showed that at the extraction wells, the distribution of peak temperatures (maximum temperatures during a heat pulse) along the screens was not uniform. Different zones of uniform peak temperatures were defined. In this way, different zones with different (uniform) inflow rates were derived using an empirical relationship between the screen-averaged peak temperature Tpk and the average flow rate of the extraction well screen normalized by the total length of the screen (Qnf). The relationship based on all well screens together showed a good fit and could be used to determine the relative contribution of each screen section. In this way, sections of higher and lower flow rates could be distinguished. Screen-specific analyses resulted in more accurate Tpk - Qnf relationships and somewhat more pronounced flow distribution variability (in line with the general relationship), but the accuracy of the empirical relationship is biased by sparse data. Moreover, all relationships have a finite applicability; at low flow rates, the Peclet number is low, such that heat transport by conduction cannot be neglected anymore. At higher flow rates, the amount of heat produced may not be sufficient to create a significantly higher peak temperature relative to the accuracy of the temperature measurements and the background temperature (low signal to noise ratio). Nevertheless, the monitoring has shown how AH-DTS can be used to monitor flow rates and infer flow rate variability within a well screen. The ERT data could be used to produce two- and three-dimensional images of the resistivity distribution, which yielded insight into the salinity dynamics during the pilot. The raw resistance data showed overall low contact resistances, indicating good overall data quality. The calculated resistivities at the shortest electrode spacing, which were used to infer two-dimensional (time-lapse) images of the resistivity dynamics along the depth of the boreholes, showed an expected influence by the backfill of the boreholes, but also a good correlation with the pumping regime and resulting salinity dynamics. The inversion results also indicated a correlation with the pumping regime, but were prone to artifacts at the depth of the brackish water extraction screens, showing amongst others an unrealistic zone of lower resistivities during Phase 1. Conclusions and recommendations Regarding the main objectives of the AH-DTS and ERT, the following can be concluded: - Three-dimensional spatiotemporal variations in salinity The raw ERT data and inversion results suggest a symmetrical salinity distribution at the scale of the inversion model during phase 1 (brackish water extraction), and an asymmetrical salinity distribution during phase 2 (freshwater extraction and saltwater upconing). A possible explanation for this, is regional groundwater flow (negligible in the saline groundwater, significant in the fresh groundwater). Other possible explanations are rotational flow due to the density differences or preferential flow related to a heterogenous subsurface. Numerical variable density groundwater flow models can be used to further investigate this. - Options for improving operational management of brackish groundwater extraction Due to the artifacts found in the inverted ERT data, no accurate prediction of the salinity of the different well screens could be made, not even when adopting different formation factors. This effect is more important than the variable inflow of groundwater into the screens as inferred from the ERT. As such, operational management of brackish groundwater extraction by automatic adjusting pumping rates to the ERT monitoring data was not further investigated. The ERT data could, however, be used to monitor the saltwater up- and downconing in close proximity to the extraction wells by using the raw (non-inverted) data, which showed a good general correspondence with the pumping dynamics. Additionally, experience with AH-DTS and ERT in this pilot has yielded the following insights: - AH-DTS monitoring has shown to be useful to 1) infer/verify backfill design of wells, 2) qualitatively determine flow rate variability along well screens, and 3) by relating average peak temperatures to the total flow rate, inferring flow rate (variability) quantitatively. - ERT data was conveniently collected, and data quality and screening results can be organized relatively straightforward. Inverting the data is much less evident, and artifacts are not easily removed automatically without additional information that can be used to guide the inversion. The main recommendations for future AH-DTS and ERT monitoring of brackish water extraction or similar processes are, apart from improvements on data acquisition and handling/transfer (grid power back-up and restart capacity, etc.): - Numerical modelling and laboratory measurements of AH-DTS to quantify the relationship between peak temperature and flow rate, to be able to quantify flow rates without knowing the total flow rate of a well. - Investigate how AH-DTS can be used to monitor process that influence inflow of groundwater into screens, such as clogging. - Using numerical variable density groundwater flow and solute transport models in combination with forward and inversion modelling of ERT to optimize the design of the ERT survey, as well as the inversion of the ERT data. - Further investigate how ERT inversion can be improved/constrained by using prior information such as geological models or other geophysical data.