building and evaluation of geologic and hydrogeological models representing saturated and unsaturated flow condi-tions. Hydrogeologists, hydrochemists, geologists, and water managers can share data and results within the same environment, which encourages collaboration between disciplines. The workflow can be used to develop and test solutions to comprehensive water management issues, with unique capabilities to address all aspects of a problem, from initial characterization and model building to simulation, monitoring, and model calibration and updating.
The workflow tools provided are scalable and therefore can efficiently incorporate as many or as few of the advanced tools available as required. As a result, the workflow is ideal for both small- and large-scale water manage-ment projects, addressing the most basic to the most complex problems. A wide variety of field data can be com-bined with models to verify correlations and validate interpretations.
The workflow allows the integration of the following disciplines:
• geophysical interpretation,
• surface imaging and mapping,
• log interpretation and well correlation,
• complex fault and fracture modeling,
• facies and geophysical modeling,
• hydrodynamic test analysis,
• uncertainty analysis,
• surface and subsurface interaction,
• upscaling processes and property population,
• flow and mass transport simulation.
Advanced tools for aquifer characterization (Schlumberger-Technoguide, 2003) allow the development and explo-ration of realistic solution scenarios, reducing uncertainty and risk. Inconsistencies that may be difficult to identify in two dimensions (2D) are immediately apparent in 3D. By discounting conceptual models that do not fit the avail-able data, uncertainty in the interpretation is reduced, resulting in a more robust model. The process is then con-cluded by the development of a 3D hydrogeological model that represents the conceptual model used to generate a 3D hydrogeological grid (gridding process).
The numerical aquifer simulator ECLIPSE (Schlumberger-Geoquest, 2003) is an integrated part of the workflow tools, providing a fully implicit, density-dependent, multiphase 3D flow and mass transport solution. The simulator is based on the proven technology of twenty years of experience as the reservoir simulation software for the oil and gas industry (Ellis et al., 1996). Aquifer systems can be more fully understood with the simulation of variably satu-rated conditions, flow and mass transport modeling, and density-dependent modeling for brine or coastal aquifers.
Processes that may influence the well performance such as clogging and flow dependent skin can be addressed. The workflow platform allows building and updating of large aquifer models in near real time.
S T O R I N G WAT E R I N U N C O N F I N E D C O N D I T I O N S
A large number of surficial aquifers in the Middle East consist of Alluvial and Eolian sediments. The total thickness of the sediments varies between 5 and 100 meters. The saturated aquifer is generally less than half of the total thick-ness of the sediments. The aquifers are relatively heterogeneous and heavily dominated by paleo-channel systems generated by surface run-off from large rainfall events through wadi systems during deposition of the alluvial sedi-ments. Native groundwater quality varies from freshwater to saline.
Storing water in unconfined conditions tends to be more complex compared to storage in confined conditions due to a number of additional considerations that have to be evaluated when investigating the aquifer. Unconfined
storage systems are more vulnerable to surface contamination because of their hydrostatic setting. Generally abstraction of groundwater is dominated by drilling hundreds of shallow wells for agricultural purposes, which obviously influence the hydrodynamic behavior of the aquifer. Sediments are unconsolidated, challenging well construction techniques to meet high well efficiencies and to control sanding issues. Generally the groundwater velocity is larger in unconfined aquifers and must be regarded as one of the critical suitability constraints. Specific yield must be taken into account to evaluate the transient behavior of the cone of depression of the wells.
Injecting high volumes of water into an aquifer will result in a mound of water (Figure 1) at the injection well. An undesired effect that develops during injection is the mound that creates a radial temporary hydraulic gradient, which allows injected water to move away from the ASR well. The fact that water raises above the static water table results in re-wetting of the unsaturated zone. However, very limited hydraulic information can be gathered from standard investigative methods to obtain unsaturated zone properties (e.g. moisture content, relative permeabilities, capillary pressure and water saturation) to predict the movement of water through this zone.
Because of well interference the maximum drawdown must be evaluated with care in large well fields. Especially in a hydraulic setting with a relatively thin saturated aquifer thickness, it may not be feasible to sustain the high recov-ery rates desired for emergency cases for long durations.
A S R W E L L F I E LD O P T I M I Z AT I O N
The operational performance of the ASR well field will subsequently reveal the actual effective recovery efficiency of the ASR system, which is used to measure the success of the ASR system (Pearce, 2001). Careful design of the well field is needed to optimize the recovery efficiency, the total recovery rate, the area of influence and the area covered by the well field. Generally geologic and hydrogeologic heterogeneities are not known at the design stage of an ASR project, which will have a large impact on the optimization.
The following factors have a key influence on the success of the well field:
• Heterogeneities of the aquifer (hydrodynamic dispersion),
• Movement of the groundwater,
• Well performance,
• Well field layout.
Practical experience reveals that unexpectedly low ASR efficiency is mainly due to the lack of full understanding of Figure 1. Cross-section showing a typical water table profile
of a surficial alluvial aquifer ASR wellfield during injection ASR wellfield
Base aquifer Initial water table Ground surface
Saturated Aquifer Unsaturated
Water table Zone
during injection
Local gradient Injection
mound
Groundwater flow
the aquifer characteristics. Macro-heterogeneities may be identified from further site investigations by means of drilling wells and/or resistivity surveys. Hydrodynamic dispersion is a particularly major unknown.
In unconfined aquifers the groundwater velocity is a major concern because the injected freshwater bubble may move away from the well field. In addition to the movement of the aquifer water, a groundwater velocity will be temporarily superimposed due to a freshwater mound during the injection period.
The well performance (efficiency of the well) is generally over-estimated during the optimization of an ASR well field using numerical simulations. In unconsolidated sand aquifers in the Middle East, clogging effects that build up over time during injection and recovery must be considered.
It has been demonstrated in the literature that the well field layout has a considerable effect on the recovery efficiency of an ASR system. Optimizing the ASR well field efficiency aims mainly to reduce the mixing that occurs between the injected water and the native groundwater in the aquifer. Mixing occurs generally at the edge of the interface between the freshwater bubble and the native water. Mixing can be reduced if the area of the interface is reduced. It is therefore desirable that the freshwater bubbles of each well in an ASR well field connect to create a single freshwater bubble at depth. If a reasonable connection of the freshwater bubbles of each ASR well is not guaranteed, saline native water may be trapped between the freshwater bubbles, having a negative impact on the recovery efficiency.
In order to design a large ASR well field, the arrangement of the wells must be optimized. The distance between ASR wells must be evaluated with great care for the following reasons:
• A large distance will reduce interference effects.
• A small distance will ensure efficient connection of the freshwater bubbles.
The objective of the optimization is to find a maximum connection of the bubbles with a minimum well inter-ference.
R E S U LT S
Injection and recovery cycles are simulated for three ASR wells that are operating with a constant rate and equal duration simultaneously. The aquifer is unconfined, homogeneous and a hydraulic gradient exists to the north-west.
The salinity and pressure of the system was monitored at the center-point of the wells for a distance of 300 m and 150 m between the wells. The simulation results are displayed for the final cycle period.
At a distance of 300 m between the three wells the freshwater bubbles will not connect (Figure 2 left) at the end of the simulated duration. At the center-point between the three wells saline native water is trapped which will impact the recovery efficiency of the ASR system.
In order to optimize the recovery efficiency the wells should be moved closer to each other. Reducing the distance between the wells to 150 m will result in a combined single freshwater bubble (Figure 2 right).
The breakthrough curve of the aquifer salinity for a single injection period at the center-point of the wells clearly displays the improved mixing that has occurred for the 150 m case (Figure 3).
Varying the distance of wells in a well field will effect the head distribution due to well interference. The
com-parison of the total head at the center-point of the wells for the two simulated cases shows a 2 m increase for the 150 m case. During recovery similar results have been achieved. The total drawdown at the center-point will be larger for the 150 m case. Figure 4 shows the dynamic behavior of the water table for an injection and recovery period crossing the well field.
Generally the maximum drawdown is of special interest to prevent any pumps to run dry in the wells.
Figure 2. Development of freshwater bubble after injection into three ASR wells spaced 300 m (left) and 150 m (right) apart
Freshwater
Figure 3. Comparison of breakthrough curves of the aquifer salinity at the center-point between the three ASR wells
Figure 4. Schematic cross-section displaying dynamic water table for injection and recovery period for 300 m and 150 m case
C O N C L U S I O N S
The recovery efficiency of an ASR well field can be optimized if mixing between freshwater and native water is minimized. This can be achieved by allowing the freshwater bubbles of each ASR well to connect to a single fresh-water bubble at depth.
This can be achieved by finding the optimum spacing for each ASR well resulting in the maximum connection of the bubbles with a minimum well interference.
By optimizing the spacing of the ASR wells to create a single water bubble, the recovery efficiency can be sub-stantially improved.
Special care must be taken to evaluate the interference between the wells that will effect the depth of the water table during injection and recovery. Well field optimization allows preventing undesirable storage effects like trapped native water and increasing drawdown causing pumps to run dry.
A new integrated aquifer characterization workflow was developed and linked with a numerical simulator that is capable of simulating complex hydrogeological conditions including variably saturated and density-dependent groundwater systems.
R E F E R E N C E S
Ellis, D.V. et al. (1996). Environmental Applications of Oilfield Technology. Oilfield Review Schlumberger, Autumn, pp. 44 – 57.
Herrmann, R., Pearce, M., Burgess, K., Priestley, A. (2004). Integrated aquifer characterisation and numerical simu-lation for aquifer recharge and storage at Marco Lakes, Florida, British Hydrological Socienty, Hydrology: Science
& Practice for the 21st Century. Vol I.
Pearce, M.S. (2001). Fundamentals of ASR Wellfield Design and Performance Evaluation. Oral presentation at the American Ground Water Trust , ASR I conference. Orlando, Fl.
Pyne, D.G. (1995). Groundwater Recharge and Wells. Boca Raton, FL: Lewis Publishers.
Schlumberger-GeoQuest. (2003). Eclipse 100 Users Manual. Abingdon, UK.
Schlumberger-Technoguide. (2003). Petrel Workflow Tools, User Manual. Oslo, Norway.
Abstract
At the Arrenæs AR trial plant, Zealand, Denmark, investigations of the geochemical and microbial processes occurring in the unsaturated zone when recharging water from Lake Arresø have been performed. Focus was on ensuring a continuously stable quality of the abstracted AR water. The investigations revealed that the primary changes in the geochemical composition and the removal of bacteria below both an irrigated area and an infil-tration basin occur in the uppermost part of the soil. The difference in the decalcification front between the irrigated and a non-irrigated area is app. 3 m. Simulations show that the calcite buffer capacity in the un-saturated zone is not depleted even in a long-term perspective, indicating that problems with aggressive CO2 should not be expected. A removal capacity at the AR plant of up to 97-99% of the bacteria was observed; how-ever, break through of coliform bacteria was seen twice in the abstracted AR water. It is estimated that only 17%
of all break through with coliform bacteria in the abstraction wells is found.
Keywords
Artificial recharge, unsaturated zone, geochemical composition, bacterial removal, filter capacity, modelling.
I N T R O D U C T I O N
Since 1995, Copenhagen Energy has run a trial plant based on artificial recharge (AR) to determine the potential for use of this technology in Denmark to ensure the supply of drinking water to Copenhagen. (Copenhagen Water Supply and County of Frederiksborg, 1995; Copenhagen Energy, 2000). The results of running the trial plant have been positive and a large-scale production plant is therefore proposed at the same location (Brandt, 1998; Hartelius et al., 2001). The Arrenæs AR trial plant consists of four infiltration basins and two trough systems (1,000 m2 each – in operation all year), and a grass-covered irrigation area (20,000 m2 – in operation from April to November), see Figure 1, (Passow, 1996). The amount of runoff water from Lake Arresø used as recharge water is app. 100,000 m3/year to the irrigation area and 300,000 m3/year to the infiltration basins and trough systems, resulting in a production of app. 270,000 m3/year, which is currently not used for drinking-water purposes. The unsaturated zone consists of glaciofluvial sand and is app. 25 – 26 m thick. Figure 1 shows a map of the area and also the water table. The hydraulic heads are affected by abstraction well I1 and an abstraction well located just north of trough system West.
With focus on ensuring a continuously stable quality of the abstracted AR water, the geochemical and micro-bial processes occurring in the unsaturated zone when
at the Arrenæs artificial recharge trial plant
T.Ø. Jensen, R.L. Berg, L. Bennedsen, G. Brandt and H. Spliid
investigated. This includes investigation of the difference in geochemical composition and bacteria content in sedi-ment samples from an irrigated area, a non-irrigated area (reference area), and an infiltration basin, and the risk of break through of pathogenic indicator organisms in an abstraction well. These investigations performed within the EU-project ARTDEMO constitute the background for establishing a management strategy including an early-warn-ing system for an AR plant.
M E T H O D S
The investigations include continuous monitoring (1994–2005) of the quality of the input water, soil water and abstracted AR water. Additionally, sediment samples from 0 – 30 m below surface (m.b.s.) were collected in October 2004 from an irrigated and a non-irrigated area using the solid flight auger drilling method. In March 2004, sediment samples were collected from an infiltration basin at depths of 0.01 (filterskin), 0.07 and 0.35 m.b.s.
Geochemical analyses performed on the soil material included pH (NEN 6411), EC (NEN 6412), non-silicate bound calcium (ICP-MS analysis on HNO3extract), calcite (Ca-extraction using HCl), organic carbon and total N (pyrolysis of decalcified sample and subsequent gas detection by LEKO analyser) and grain size distribution (< 2mm, by a FRITSCH Laser Particle Sizer A22). Chemical modelling (PHREEQC) was done using chemical parameters analysed in water samples according to the Danish Standards. Microbiological analyses of both water and sediment samples included colony-forming units (CFU) at 22°C and 37°C (DS/EN ISO 6222), coliform bacteria and thermotolerant coliform bacteria (DS 2255), sulphite-reducing clostridia (NMKL 56) and Clostridium per-fringens (DS 2256). Enterococci were analysed in both water (ISO 7899/2 mod.MST98) and sediment samples (DS 2401). The total amount of bacteria in sediment samples from the irrigated and non-irrigated area was deter-mined using the Acridine Orange Direct Count (AODC).
R E S U LT S
The geochemical results from sediment samples (0 –30 m.b.s.) collected at the irrigated and non-irrigated area at the Arrenæs AR trial plant show that the difference in geochemical properties between the two areas is significant in only the upper 2 m of the unsaturated zone, see Figure 2. In the deeper parts from 2–26 m.b.s., the geochemical properties are similar for the two areas. The water table is located in app. 26 m.b.s., which is reflected in the geo-chemical properties.
Accumulation of organic carbon is occurring in the topsoil, see Figure 2. Due to bioturbation and degradation, the content of organic carbon is decreasing in the upper 2 m.b.s. The degradation of organic carbon is reflected in a decreased pH compared to the deeper layers, where pH is controlled by calcite equilibrium.
C-org [%wt / wt)
Figure 2. Organic carbon, pH, calcium, and cation exchange capacity (CEC) at the irrigated and the non-irrigated area
Data show a significant increase in the content of calcium from 0–2 m.b.s. at the irrigated area. This calcium is primarily present as calcite. From the cation exchange capacity, Figure 3, it is estimated that ion exchange does not have a significant influence on the calcium content in the sediment. The accumulation of calcite in the upper 2 m is primarily a result of irrigation with lake water supersaturated with calcite, and evaporation. In addition, algae growth in the topsoil and loss of CO2gas may cause precipitation of calcite. Figure 2 indicates a decalcification front located app. 1–2 m.b.s. at the non-irrigated area and below the calcite rich soil horizon in app. 3–4 m.b.s. at the irrigated area. Whether this difference can be explained by artificial recharge needs further investigations of the location of the decalcification front, as natural geological heterogeneities also have an impact on the difference shown in Figure 2. Below the leaching front, geological variations may influence the carbonate equilibrium.
Similar investigations were performed in the upper 0.35 m of the soil below an infiltration basin. Figure 3 shows the content of calcite and organic carbon together with electrical conductivity. The high calcite content in the filter-skin (0.01 m.b.s.) may be explained by algae growth in the infiltration basin causing a decrease in CO2and thus increasing the calcite saturation index (Schuh, 1990). The organic carbon content in the sediment below the infil-tration basin decreases from 0.095 to 0.031% in the upper 0.07 m of the soil, see Figure 3. The degradation of organic carbon is reflected in a concurrent decrease in calcite content and electrical conductivity.
The geochemical processes in the unsaturated zone may also be affected by complicated flow patterns such as pref-erential flow, as revealed in studies at the Arrenæs AR trial plant by Brun and Broholm (2001).
The microbiological results from the sediment samples collected at the irrigated and non-irrigated area (0 –30 m.b.s.) indicate that bacteria are primarily present in the upper 2 m, see Figure 4.
CaCO3 [mg/kg]
Figure 3. Calcite, organic carbon and electrical conductivity (EC) in the soil below an infiltration basin (n = 2)
AODC [cells/g]
Figure 4. AODC, CFU, coliform bacteria, thermotolerant coliform bacteria, sulphite-reducing clostridia (SRC) and Clostridium perfringensfrom 0–5 m.b.s. at the irrigated and non-irrigated area
CFU at 22°C and 37°C detected using the yeast extract agar are higher in the upper 2 m at the non-irrigated than at the irrigated area. The total number of bacteria (AODC) is similar at the two locations, indicating the presence of bacteria cultures not accounted for using the yeast extract agar. As expected, the content of patho-genic indicator organisms is generally higher at the irrigated than at the non-irrigated area. Sulphite-reducing clostridia and Clostridium perfringens were present at 0 –2 m.b.s. at both the irrigated and non-irrigated area, whereas coliform bacteria, thermotolerant coliform bacteria, and enterococci only were detected at the irrigated area from 0 – 2 m.b.s., Figure 4. As a total, the results show a reduction of up to 99% of the pathogenic indicator organisms in the upper 2 m.
The studies also included analysis of sediment samples from in an infiltration basin. The results show a large reduction of bacteria in the upper 0.35 m (see Figure 5), where the content of coliform bacteria and thermotolerant coliform bacteria is reduced by up to 97%. This is similar to findings at the Arrenæs AR trial plant by Jørgensen (2001).
The studies also included analysis of sediment samples from in an infiltration basin. The results show a large reduction of bacteria in the upper 0.35 m (see Figure 5), where the content of coliform bacteria and thermotolerant coliform bacteria is reduced by up to 97%. This is similar to findings at the Arrenæs AR trial plant by Jørgensen (2001).