muFlowReacT: A Library to Solve Multiphase Multicomponent Reactive Transport on Unstructured Meshes.

In this paper we present a new reactive transport code for the efficient simulation of groundwater quality problems. The new code couples the two previously existing tools OpenFoam and PhreeqcRM. The major objective of the development was to transfer and expand the capabilities of the MODFLOW/MT3DMS-family of codes, especially their outstanding ability to suppress numerical dispersion, to a versatile and computationally efficient code for unstructured grids. Owing to the numerous, previously existing transport solvers contained in OpenFoam, the newly developed code achieves this objective and provides a solid basis for future expansions of the code capabilities. The flexibility of the OpenFoam framework is illustrated by the addition of diffusional processes for gaseous compounds in the unsaturated zone and the advection of gases (multiphase transport). The code capabilities and accuracy are illustrated through several examples: (1) a simple 2D case for conservative solute transport under saturated conditions, (2) a gas diffusion case with reactions in the unsaturated zone, (3) a hydrogeologically complex 3D reactive transport problem, and finally (4) the injection of CO2 into a deep aquifer with acidification being buffered by carbonate minerals.

Identifying remedial solutions through optimal bioremediation design under real-world field conditions.

Over more than a century of intense industrial production and associated accidental release, petroleum products (e.g., gasoline, diesel, fuel oil) have contaminated a significant portion of the world's groundwater resources. Groundwater remediation is generally a complex task, especially where aquifers and the associated contaminant distribution are highly heterogeneous. The ability to predict the efficiency of such remediation is of crucial importance, as the costs are strongly linked to the treatment design and duration. In this study, a coupled simulation-optimization (S/O) framework, consisting of a process-based reactive transport simulation model linked with particle swarm optimization (PSO) was developed. It was subsequently applied for the design of a real-world in situ bio-treatment of a BTEX contaminated aquifer in France. In the application, the optimization framework was used to simultaneously determine optimal well locations and their optimal injection rates, both constituting key elements of the enhanced biodegradation design problem. The optimization of the treatment efficiency was examined in terms of three different regulatory objectives, (1) minimization of the residual NAPL mass of the key contaminant, i.e., benzene, in the source zone, (2) reduction of the maximum concentration of benzene in groundwater, and (3) minimization of the time required to reduce the benzene concentration in groundwater to below a threshold value. Our analysis of potential, optimal remediation strategies showed that: (i) the complexity of the biodegradation behavior at real sites may favor very different remediation options as a result of varying remediation targets, (ii) the long term behavior of the contaminants after the end of the active treatment period, which is often neglected, showed to have a significant influence on remediation design that requires increased attention, (iii) PSO has shown to be a very efficient algorithm in the context of the present study. The insights that can be gained from such a framework will provide decision support to select the most suitable remediation strategy while facing different regulatory objectives.

An iterative strategy for contaminant source localisation using GLMA optimization and Data Worth on two synthetic 2D Aquifers.

A contaminant source localisation strategy was developed considering unknown heterogeneous hydraulic conductivity field, unknown dispersivity and unknown location of a continuous contaminant source. The Gauss-Levenberg-Marquardt algorithm is combined with a data worth analysis to estimate the unknown parameters and identify the best locations of additional measurements. The data collection strategy is iterative, based on the ability of the additional dataset to decrease the uncertainties on the contaminant source location. Two 2D synthetic models are considered. The method is first illustrated with a simple model and a more complex model is then considered to evaluate the ability of the approach to locate the contaminant source from hydraulic heads and concentration data. This approach is parsimonious in terms of model runs and applicable to real cases. The results give a good estimate of the source location and the dispersivity, with acceptable NRMSE for each case. New observations introduced at each iteration decrease the standard deviation of the source location and improve the NRMSE. The estimated hydraulic conductivity field presents the same features as the original field.

On the influence of groundwater table fluctuations on oil thickness in a well related to an LNAPL contaminated aquifer.

This paper presents a new modelling approach to describe and explain the temporal variation of oil thickness in well due to groundwater table fluctuations. This new model, which intends to be simple and easy to implement, was compared to field data obtained by continuous measurements of vertical LNAPL position in wells. Two scenarios have been studied: a pumping well where the oil layer is unconfined, and one where the oil layer is present in a confined porous media. This study shows that the time-depend fluctuation of the oil thickness observed in the wells could not be reproduced only with the differences between the residual oil saturations (Sorw and Sora) as suggested by Kemblowski and Chiang (1990). It should consider the transient mass exchange between the well and the porous media. Also, the proposed model shows that making the assumption of equilibrium conditions as suggested by Lenhard et al. (2017) for calculating the volume exchanges between the wells and its surrounding introduced errors. Considering transient transfers of oil better reflects the field observations. This observation is a key outcome for improving field data interpretation (e.g.: bail-down test data) and the remedial approach at site polluted by mineral oils.

Removal of NAPL from columns by oxidation, sparging, surfactant and thermal treatment.

In this paper, four treatment techniques commonly applied to Volatile Organic Compounds (VOC) removal from soil are compared in column experiments with pure sand containing a residual Light Non-Aqueous Phase Liquid (L-NAPL) contamination. Oxidation is tested through the injection of Fenton reagent, with persulfate, and combined with sparging with the injection of ozone. Surfactant treatment was conducted at low flow rates with Tween®80. Sparging was conducted by air injection but at a low flow rate of 1 mL min-1. Finally several columns were thermally treated at a temperature of 80 °C. The results showed high removal (>90%) for all techniques used, although only thermal treatment on BTEX (Benzene, Toluene, Ethylbenzene and Xylenes) reached 100% efficiency. The main limiting factors of each technique were: (i) for oxidation, the solubility of the substance limited the removal; (ii) for surfactant both the solubility in the surfactant and the type of surfactant are important; (iii) for sparging, the main factors are contaminant vapor pressure and porous media grain size; (iv) for thermal treatment, the limitation arises from the contaminant vapor pressure and the medium hydraulic conductivity. A comparison with literature data shows that the results are consistent with most of the studies conducted on one technique.

Comparison of residual NAPL source removal techniques in 3D metric scale experiments.

This study compared four treatment techniques for the removal of a toluene/n-decane as NAPL (Non Aqueous Phase Liquid) phase mixture in identical 1 cubic meter tanks filled with different kind of sand. These four treatment techniques were: oxidation with persulfate, surfactant washing with Tween80®, sparging with air followed by ozone, and thermal treatment at 80°C. The sources were made with three lenses of 26×26×6.5cm, one having a hydraulic conductivity similar to the whole tank and the two others a value 10 times smaller. The four techniques were studied after conditioning the tanks with tap water during approximately 80days. The persulfate treatment tests showed average removal of the contaminants but significant flux decrease if density effects are considered. Surfactant flushing did not show a highly significant increase of the flux of toluene but allowed an increased removal rate that could lead to an almost complete removal with longer treatment time. Sparging removed a significant amount but suggests that air was passing through localized gas channels and that the removal was stagnating after removing half of the contamination. Thermal treatment reached 100% removal after the target temperature of 80°C was kept during more than 10d. The experiments emphasized the generation of a high-spatial heterogeneity in NAPL content. For all the treatments the overall removal was similar for both n-decane and toluene, suggesting that toluene was removed rapidly and n-decane more slowly in some zones, while no removal existed in other zones. The oxidation and surfactant results were also analyzed for the relation between contaminant fluxes at the outlet and mass removal. For the first time, this approach clearly allowed the differentiation of the treatments. As a conclusion, experiments showed that the most important differences between the tested treatment techniques were not the global mass removal rates but the time required to reach 99% decrease in the contaminant fluxes, which were different for each technique.

Elucidating the fate of a mixed toluene, DHM, methanol, and i-propanol plume during in situ bioremediation.

Organic pollutants such as solvents or petroleum products are widespread contaminants in soil and groundwater systems. In-situ bioremediation is a commonly used remediation technology to clean up the subsurface to eliminate the risks of toxic substances to reach potential receptors in surface waters or drinking water wells. This study discusses the development of a subsurface model to analyse the performance of an actively operating field-scale enhanced bioremediation scheme. The study site was affected by a mixed toluene, dihydromyrcenol (DHM), methanol, and i-propanol plume. A high-resolution, time-series of data was used to constrain the model development and calibration. The analysis shows that the observed failure of the treatment system is linked to an inefficient oxygen injection pattern. Moreover, the model simulations also suggest that additional contaminant spillages have occurred in 2012. Those additional spillages and their associated additional oxygen demand resulted in a significant increase in contaminant fluxes that remained untreated. The study emphasises the important role that reactive transport modelling can play in data analyses and for enhancing remediation efficiency.

Comparison of soil solution sampling techniques to assess metal fluxes from contaminated soil to groundwater.

The unsaturated zone plays a major role in elemental fluxes in terrestrial ecosystems. A representative chemical analysis of soil pore water is required for the interpretation of soil chemical phenomena and particularly to assess Trace Elements (TEs) mobility. This requires an optimal sampling system to avoid modification of the extracted soil water chemistry and allow for an accurate estimation of solute fluxes. In this paper, the chemical composition of soil solutions sampled by Rhizon® samplers connected to a standard syringe was compared to two other types of suction probes (Rhizon® + vacuum tube and Rhizon® + diverted flow system). We investigated the effects of different vacuum application procedures on concentrations of spiked elements (Cr, As, Zn) mixed as powder into the first 20 cm of 100-cm columns and non-spiked elements (Ca, Na, Mg) concentrations in two types of columns (SiO2 sand and a mixture of kaolinite + SiO2 sand substrates). Rhizon® was installed at different depths. The metals concentrations showed that (i) in sand, peak concentrations cannot be correctly sampled, thus the flux cannot be estimated, and the errors can easily reach a factor 2; (ii) in sand + clay columns, peak concentrations were larger, indicating that they could be sampled but, due to sorption on clay, it was not possible to compare fluxes at different depths. The different samplers tested were not able to reflect the elemental flux to groundwater and, although the Rhizon® + syringe device was more accurate, the best solution remains to be the use of a lysimeter, whose bottom is kept continuously at a suction close to the one existing in the soil.

On the role of low-permeability beds in the acquisition of F and SO4 concentrations in a multi-layer aquifer, South-West France.

Fluoride (F(-)) commonly threatens groundwater quality. This is the case around the city of Bordeaux (France), where numerous wells tapping the thick and complex Eocene aquifer are contaminated by fluoride, which presents an issue for drinking water supply. The joint analysis of the spatial distribution of fluoride with other species like sulfate suggests that concentrations are mainly related to the occurrence of low-permeability layers containing evaporites or fluorite deposits. In order to identify the origin of the observed concentrations, a radial flow and transport model is implemented at the borehole scale. The hydraulic conductivity of the low-permeability layers and the vertical dispersivity of the aquifer were optimized to match the observed values of sulfate and fluoride concentrations. Interestingly, each of these parameters influences differently the simulated concentrations. This model has been successfully implemented to a neighboring well with the same parameter values, which tests the approach. The major conclusions drawn are: (i) the contamination in fluoride originates from the low-permeability layers, (ii) every low-permeability layer intercepted by the well releases fluoride (iii) Contamination not only originates from pore water of low-permeability layers, but may persist with long-term pumping due to mineral dissolution. As a consequence, fluoride contamination is likely to persist for a long time and the only solution to reduce fluoride concentration in abstracted water is to seal well screens facing low-permeability layers.

Validation of a multilevel sampling device to determine the vertical variability of chlorinated solvent in a contaminated aquifer.

The vertical heterogeneity of contaminant concentrations in aquifers is well known, but obtaining representative samples is still a subject of debate. In this paper, the question arises from sites where numerous fully screened wells exist and there is a need to define the vertical distribution of contaminants. For this purpose, several wells were investigated with different techniques on a site contaminated with chlorinated solvents. A core-bored well shows that a tetrachloroethene (PCE) phase is sitting on and infiltrating a less permeable layer. Downstream of the cored well, the following sampling techniques were compared on fully screened wells: low flow pumping at several depths, pumping between packers and a new multilevel sampler for fully screened wells. Concerning low flow rate pumping, very low gradients were found, which may be due to the existence of vertical flow inside the well or in the gravel pack. Sampling between packers gave results comparable with the cores, separating a layer with PCE and trichloroethene from another one with cis 1,2-dichloroethene and vinyl chloride as major compounds. Detailed sampling according to pumped volume shows that even between packers, cleaning of the inter-packer volume is necessary before each sampling. Lastly, the proposed new multilevel sampler gives results similar to the packers but has the advantages of much faster sampling and a constant vertical positioning, which is fairly important for long-term monitoring in highly stratified aquifers.

Comments on "Analytical modelling of fringe and core biodegradation in groundwater plumes." by Gutierrez-Neri et al. in J. Contam. Hydrol. 107: 1-9.

In this comment, we revisit equations concerning the analytical solutions presented by Gutierrez-Neri and co-workers for reactive transport for a pollutant undergoing core and fringe degradations. We state that a correction needs to be made in Eq. (9) of the work of Gutierrez-Neri et al. in order that the equation follows closely previous work published by J. Bear (in 1-D) and P.A. Domenico (in 3-D). Furthermore we derive alternative solutions for Eqs. (13)-(16) which separate more clearly the first-order reaction and the instantaneous reaction. It is shown that the corrected solution agrees better with the results from the numerical model than the previous solution. An improvement is also made by giving a solution which avoids negative concentrations. Furthermore, the corresponding solution for the electron acceptor reacting with the pollutant is given.

Validation of reactive model assumptions with isotope data: application to the Dover case.

This paper presents a new approach to compare simulations from reactive transport models. We show that where several compounds follow a reaction chain and can be degraded under different redox conditions, two models with significantly different parameters can lead to similar species distributions. The studied case is the aquifer contamination at Dover, DE U.S., where well characterized plumes of PCE, TCE, cDCE, and VC have been simulated both with RT3D and by a semianalytical procedure proposed here. In order to validate the models, we use isotope data of PCE and TCE that where also measured on that site. A simple calculation, validated against a geochemical model, allows us to simulate the spatial distribution of the isotopic ratio. Despite the similar spatial distribution of contaminant concentrations, simulated isotopic ratios are completely different. Field data correspond to a very localized degradation of PCE and TCE compounds. The presence of a small degradation zone has important consequences on the future evolution of the plume and on the potential remediation techniques.

Factors controlling BTEX and chlorinated solvents plume length under natural attenuation conditions.

Natural attenuation is presently used at numerous sites where groundwater is contaminated. In order to simulate this attenuation, reactive transport models are often used but they are quite complex and depend on both physical and chemical conditions in the aquifer. As complex numerical models cannot be used to study all possible cases, we develop here analytical solutions to draw general conclusions. Our strategy, called MIKSS (Mixed Instantaneous and Kinetics Superposition Sequence), allows the calculation of the concentrations of all reacting substances in a plume. It is an extension of the superimposition principle that is able to treat the case of joint kinetics and instantaneous reactions. The basic equations have been extended to treat different reactions that occur in the plume core and at its fringe. At first we consider one organic substance degraded under all oxidising conditions (toluene for instance). For this problem the size of the plume depends on the reduced source width and on the ratio of the organic substance concentration to the sum of the electron acceptors' concentrations. For several BTEX substances having different degradation behaviour the formulation is similar, but leads to quite different plume lengths for each substance. Contrary to the case of one substance, the plumes can be quite long and may not satisfy the target risk level. For chlorinated solvents we developed a specific approach to take under consideration all reactions and particularly the competition for hydrogen. A formula is given to assess the size of the plume core, i.e. the zone with highly reducing conditions. The factors influencing the core length are the same as for BTEX (source width, dispersivity, organic carbon content). The size of the TCE plume is calculated from the plume core length and the kinetic constant of TCE degradation. Using assumptions of degradation constants for DCE and VC it is also possible to calculate the longitudinal concentration profile of these substances. The degradation of moderately substituted solvents under oxic conditions reduces the size of their plumes but under these conditions TCE becomes the major threat. Among the conditions studied in this paper, very few chlorinated solvents sites can lead to a negligible risk at an acceptable distance from the source.

Kinetics of natural attenuation of BTEX: review of the critical chemical conditions and measurements at bore scale.

This paper describes the chemical conditions that should favour the biodegradation of organic pollutants. Thermodynamic considerations help to define the reaction that can occur under defined chemical conditions. The BTEX (benzene, toluene, ethylbenzene, and xylene) degradation is focused on benzene, as it is the most toxic oil component and also because it has the slowest degradation rate under most field conditions. Several studies on benzene degradation allow the understanding of the basic degradation mechanisms and their importance in field conditions. The use of models is needed to interpret field data when transport, retardation, and degradation occur. A detailed comparison of two existing models shows that the limits imposed by oxygen transport must be simulated precisely to reach correct plumes shapes and dimensions, and that first-order kinetic approaches may be misleading. This analysis led us to develop a technique to measure directly biodegradation in the field. The technique to recirculate water at the borehole scale and the CO2 analysis are depicted. First results of biodegradation show that this technique is able to easily detect the degradation of 1 mg/l of hydrocarbons and that, in oxic media, a fast degradation rate of mixed fuel is observed.

Aggregation rates of natural particle populations.

In this paper an experimental approach of aggregation in natural suspensions is presented. The suspensions are organic-matter-rich waters sampled in a brook which drains peat areas. The aggregation was conducted on raw samples in three different experiments lasting from 2 to 8 days. The particle size distribution (PSD) in the 0.5-10 microm size range was followed with a laser sizer and appeared to be almost constant along the whole experiment duration. Nevertheless, the volume of particles larger than 10 microm increased steadily, showing that aggregation occured. This appeared to be the consequence of a steady-state aggregation which allowed the removal of the whole particle set within a day. The use of an aggregation model adapted to calculations on PSD allowed estimation of the aggregation efficiency for such suspensions.

Geostatistical analysis of soil contamination in the Swiss Jura.

The topsoil of a 14.5 km(2) region of the Swiss Jura has been surveyed to identify the distributions of trace metals in it. The soil was sampled at 366 sites selected by combining a square grid and nesting. Concentrations of seven potentially toxic metals, namely Cd, Co, Cr, Cu, Ni, Pb and Zn, were measured. Land use and geology (stratigraphy) were also recorded. Variograms were bounded in the range from 110 m to 1500 m with contributions to the variance at all distances exceeding 6 m. The variograms of Cd, Cr, Cu and Pb are dominated by short range correlation, those of Co and Ni by correlation of long range, and Zn is intermediate. The concentrations were estimated at the nodes of a fine grid by ordinary block kriging and then contoured to produce maps. The maps of Co and Ni have a coarse patchy pattern similar to that of the geology, suggesting that these metals derive from the bedrock. This is supported by analysing the variance by geology. Copper and Pb have finer patterns of distribution, and are more likely to have been added with fertilizer or manure or domestic waste. Cadmium could originate from human activities, such as smelters or fertilizer spreading, or from specific geological deposits, such as moraine.