Background / objective
At a former tar production plant of ca. 6,8 ha, high concentrations of organic contaminants are present in the quaternary layer. Tar DNAPL is present at a depth of ca. 15 m-below ground level (bgl), on top of the Tertiary layers underneath important parts of the site. The contamination has migrated under the adjacent River Scheldt underneath an ecological valuable area. The DNAPL-contamination threatens the underlying Tertiary aquifer (important source of drinking water extraction) and downstream areas. The design of a remediation strategy was challenging, due to the size and depth of the contaminated area, the complex alluvial (hydro)geology, the presence of an active chemical production site and the presence of many stakeholders.
Approach / Activities
Initial feasibility studies showed that complete cleanup of the site was not cost effective. Hence, the remediation strategy design focused on elimination of the migration risk. To be able to identify the crucial aspects of the migration risk an elaborate Conceptual Site Model has been developed. In addition to conventional drillings and monitoring wells, next generation techniques have been used to map geology and contamination in three dimensions: LIF-CPT (ROST), electrical CPT and geoelectrical tomography. These data provided the means to identify gaps in the underlying Tertiary clay layer, to map the migration of DNAPL underneath the River Scheldt, to delineate in detail the presence of contamination in the different soil layers and to set up a hydrogeological model.
Results/outcomes
Based on these insights, a robust containment strategy was designed and approved by the authorities. Isolation of the contamination is to be attained by a three-sided bentonite slurry wall, to be installed in 2016/2017, combined with limited pumping of groundwater to compensate for ground- and rainwater inflow. Forty DNAPL-recovery wells are installed in some selected critical locations and at present manual pumping of DNAPL using a peristaltic pump is undertaken. A detailed testing phase provided the ideal pumping rates and tube diameters for the selective removal of DNAPL. Results showed that it was crucial to develop a different pumping schedule for every well, because of heterogeneity of the DNAPL inflow rate. First results of the full scale system are promising: a total removal rate of more than 1.000 liter DNAPL/day was attained. A further elaboration of the technique, as well as automation of the system is expected to increase the removal rate in the near future. More results will be available at the AquaConSoil conference.
Numerical contaminants’ fate and transport (F&T) modelling has been commonly used: (i) to describe and predict contaminants’ distribution and migration in groundwater, (ii) to improve understanding of processes occurring in the subsurface and influencing the contaminants’ migration; (iii) as a framework for data integration and interpretation; (iv) to assess risk associated with groundwater contamination, (v) to predict contaminants’ behavior under different scenarios, and (vi) to improve groundwater monitoring network. In this work we applied F&T modeling for selecting effective remediation of contaminated groundwater.
Selection of the most effective remediation scenario for contaminated groundwater is a complex issue that requires a suitable methodology. An important part of it is F&T modeling, preceded by field and laboratory investigations. The simulations enable to forecast contaminants’ migration in the aquifer depending on different remediation scenarios and present the results in a friendly way by visualization of concentration changes in time and space. However, the model’s credibility depends strongly on series of conditions including: proper site investigations, relevant conceptual site models, appropriate estimates for model parameters and selecting the most suitable algorithms to fit the conceptual models. The obtained results should be always treated as an approximation what is associated with the nature of the modeling tool: developing the conceptual scheme requires simplification of real-world conditions due to capability of computer programs. An insufficient access to the aquifer is another important issue that may strongly affect the modelling results. The modeling approach should include detailed calibration to match the simulated and observed parameters, and validation – the process of evaluating and testing the different aspects of the model in order to refine and enhance the model.
A site with TCE and PCE contaminated groundwater located in south-east Poland was selected as a case study to present an application of F&T numerical modeling for selecting an effective remediation strategy. A methodology was developed, that included: field and laboratory investigation followed by numerical groundwater flow modeling and contaminants’ F&T modeling. The selection of effective scenarios was based on a multi-criteria analysis (MCA), and a monitoring network was designed to control the performance of selected remediation scenario.
The contaminants’ F&T model was developed using Visual MODFLOW software in MT3D environment. Based on the defined criteria simulations were conducted for a number of selected groundwater remediation strategies: (i) natural attenuation (NA) based remediation, (ii) ‘pump and treat’, (iii) permeable reactive barriers (PRB), and (iv) in situ chemical oxidation (ISCO) applied locally in the contamination source area. To perform a comprehensive evaluation simulations were conducted for a 100-year period. Changes in contaminants’ concentrations were evaluated in 1-year periods and presented graphically. Localization, configuration and other characteristics (e.g. pumping rates, PRB permeability parameters) of the remedial installations were adjusted manually by the trials and errors method.
The estimated times required to reach the defined remediation goal (i.e. TCE and PCE concentrations in water <50 µg/dm3 required for a ‘good’ chemical status of groundwater) for studied scenarios varied from 20 to more than 100 years. The most effective remediation strategy was selected based on the MCA which included: (i) the environmental criterion (i.e. reaching the ‘good’ chemical status of groundwater: ia – in immediate surroundings of waterworks, ib – in whole study area); (ii) the waterworks operational criterion (i.e. drinking water standards (ƩTCE+PCE <10 µg/dm3); (iii) the economic criterion (investment and operational costs). Each adapted criterion was weighted based on the site-specific factors as: 0.4, 0.3, 0.2, and 0.1, respectively. The results indicated that ‘pump-and-treat’ was the most effective method of groundwater remediation under the assumed conditions, followed by PRB and ISCO. NA turned to be the least suitable requiring the longest time to obtain the remediation goal, particularly due to small sorption and unfavorable hydrogeochemical conditions for intrinsic biodegradation.
Concluding, numerical contaminants’ F&T modeling allows simulating different remediation strategies, thus it can facilitate the decision making process. The developed methodology for assessing effective remediation methods based on F&T modelling and the MCA may be applied also at other sites contaminated with other contaminants. However, the selection criteria should be set for each site individually. The prescribed weights in MCA allow to point out the most important factors to deal with from an investor’s viewpoint, as well as from administrative, formal, environmental, social, etc. perspectives.
Introduction.
In situ chemical reduction (ISCR) is a technique commonly used to transform highly toxic and soluble hexavalent chromium (CrVI) into relatively low toxic and less mobile trivalent chromium (CrIII). There have been several studies focussed on CrVI contaminated groundwater, but in situ remediation techniques of unsaturated soils have not been thoroughly tested. The Spinetta Marengo (SM) industrial site (Alessandria, Italy) was contaminated with cromium (Cr) as a result of the activities of the chrome planting facility that was dismantled in the mid-1970s. Data indicated that the source of groundwater contamination was CrVI in the unsaturated soils. A novel pilot test study was conducted at the SM site to evaluate how effective different treatment techniques were in reducing CrVI concentrations in the unsaturated zone.
Material and Methods.
Core samples were submitted for preliminary laboratory treatability studies (batch and column tests). Results indicated that sodium dithionite was the most appropriate reagent for treating the subsurface contamination.
Field scale pilot tests were conducted that involved injections of sodium dithionite in an area historically used for dichromate ore processing where some of the highest solid phase CrVI concentrations had been observed in the unsaturated zone.
Pre-treatment sampling. Soil samples were collected to the maximum depth of 6m bgl and analysed for solid phase total concentrations of Cr, CrVI, Fe, S, Mn, As, Ni and Fraction Organic Carbon (fOC). The CrVI concentrations ranged from 78 to 3000 mg/kg. The monitoring network implemented for the pilot test consisted of three fully screened piezometers and six boreholes for georadar investigation. Pre-treatment CrVI concentrations in groundwater ranged from 0.45 to 3.2 mg/L.
A stoichiometric excess of sodium dithionite was added to an aqueous solution with a sodium carbonate pH buffer.The solution was formulated and immediately injected to prevent dithionite reaction with O2. The solution presented a negative redox potential (-600 mV vs NHE), a significant electrical conductivity, and alkaline pH (10-12). Three injection tests were conducted from June to October 2012.
Injection Test I consisted of a grid with four direct push injections, spaced 1.8 m from one another. At each point, 1000 L of solution were delivered between 2.0-4.0 m bgl, at a flow rate of 70 L/min. The direct push rod was fitted with a four-port Geoprobe® injection tip to facilitate reductant delivery at specified depths.
Injection Test II included two injection wells installed 1.8 m from one another. The two wells consisted of two inch diameter PVC screened from 2.0-4.0 m bgl. At each point, 1000 L of solution was injected at an average rate of 125 L/min.
Injection Test III included a single multi-step direct push injection. During injection, a 500 L solution was delivered between 0.5-1.0 m bgl, to force the reductant into the filling material and fine-grained silt unit, and an additional 2000 L solution was delivered between 2-4 m bgl.
Geophysical monitoring. High resolution electrical resistivity tomography, surface and borehole georadar techniques were performed before, during and after the injections, and a time-lapse geophysical approach was adopted.
Groundwater monitoring. Based on numerical modelling simulations, a groundwater monitoring campaign was carried out to determine baseline hydrochemistry and hydraulic heads and continued during and after the injections with defined intervals, approximately ranging from 24 hours to a week. All the groundwater samples were analysed on site for CrVI, electrical conductivity, temperature and pH. Additional groundwater samples were collected and analysed for total concentrations of Cr, CrVI, Mn, As, Ni, Fe, SO42-.
Post-treatment soil sampling. Based on geophysical evidences, post-treatment soil samples were collected in the highly perturbed areas (resistivity anomalies lesser than -30%) to evaluate the effectiveness of the treatment.
In situ water flushing. An injection of 1500 L of water into the treatment zone was accomplished with Geoprobe® technology in November 2012. Groundwater samples were collected and analysed before, during and after injection to verify the mobility of CrVI in the unsaturated soils.
Results
The injection caused a significant reduction of resistivity in the unsaturated soils, ranging between -10% and -70%. The data indicated an average radius of influence of 1.1 m for each single direct push injection, with a maximum of 1.5 m, and a treated unsaturated soil volume ranging from 6-9 m3. Post-treatment soil samples collected in the treatment zone indicated a significant decrease in CrVI concentrations: less than the detection limit of 1 mg/kg. As expected, the excess of dithionite rapidly turned to sulphate and insoluble metal sulfides were formed. The total sulfur concentrations in the unsaturated soils increased to 3490 mg/kg following the reductant injection. The average total Cr concentrations in the core samples collected before and after treatment showed no significant changes.
Conclusions.
The investigations proved the capacity of geophysical monitoring to clearly evidence the reductant migration in the unsaturated zone. Concentrations of CrVl in the unsaturated soils were reduced by approximately 80-100% after injecting sodium dithionite with a four-port Geoprobe® tip. Finally, pilot testing indicated that direct push injections of sodium dithionite at specified depths is a viable treatment technology for unsaturated soil remediation at the SM site.
Acknowledgments. The authors would like to thank Solvay Specialty Polymers for funding this project.
Background/Objectives: A combination of laboratory experiments, field hydraulic and transport tracer studies, and numerical modeling were used to design a Surfactant Enhanced Aquifer Remediation (SEAR) of a sandy aquifer contaminated by petroleum based jet fuel (LNAPL) at a Danish Defense tank storage facility situated in western Jutland in Denmark. In the full scale application, a blend of anionic surfactant in a brine solution (sodium chloride) will be injected into the subsurface to mobilize free-phase and residual NAPL from subsurface soils to downstream extraction wells. The SEAR will utilize foam in the aquifer to improve mobility control and to enhance remediation in the smear zone. The use of surfactant and foam is a well-established technology for enhanced oil recovery in the oil industry. The technology has also been proven in remediation of both LNAPL and DNAPL in both the laboratory and in the field. The use of numerical models, calibrated to both laboratory and field studies, at every stage of the design process has provided essential information for the design of the final SEAR, currently scheduled for spring 2015.
Approaches/Methods: During planning and design of the SEAR, numerical models have been tools for understanding phase behavior, decisions on use of technologies, and in the SEAR design. As a first step, the multi-phase model UTCHEM was calibrated to laboratory partitioning experiments to ensure that phase behavior could be accurately simulated over the range of expected field conditions. UTCHEM phase behavior and other physical/chemical parameters were further refined by matching simulated breakthrough curves of aqueous, oil, and microemulsion phases to breakthrough curves measured in laboratory column studies of surfactant flooding. Very close matches between measured and simulated phase behavior and breakthrough curves were obtained.
The field-scale SEAR design was completed in several stages. First, the numerical models MODFLOW and MODPATH were used to design a preliminary injection/extraction well network. The MODFLOW model hydraulic conductivities, storage coefficients, and stratigraphy were adjusted until the flow patterns matched data from pumping and slug tests. A chloride tracer test was then simulated with the MODFLOW and MT3D to estimate the degree of subsurface heterogeneity, fine-tune flow and transport parameters, and ensure that injected fluids would be captured by the extraction wells.
Following the hydraulic simulations, a UTCHEM model was created that matched the characteristics of the MODFLOW model. The UTCHEM model was then run to determine optimal fluid injection rates, expected phase production at the extraction wells, injection and production well depths, foam injection rates and effects, and other operating parameters. The operating parameters and alternative well configurations were altered to maximize oil removal while ensuring recovery of injected fluids. The final UTCHEM simulations form the basis for the planned SEAR operating conditions
Applications: The combination of multi-phase simulations, field studies, and laboratory experiments is a powerful technique for designing SEAR remediation projects. SEAR projects frequently do not meet performance expectations because the complexities of the processes involved are difficult to estimate without careful study. These design techniques can be applied to many other planned SEAR projects to increase their chances for success and give regulators better data on which to base decisions.
Lessons learned: The use of numerical models has been a helpful tool in understanding the remediation processes compared to the often used “black box” approach, in which remediation selection and performance are evaluated by laboratory parameters and monitoring data from the field without a thorough understanding of the processes. The correlation between laboratory, field and model results gave understanding, knowledge and credibility in choice of surfactant blend and set up and design of the full scale remediation. The results of the simulations were also helpful to regulators, who were able to use the results to set up terms and permits for the full scale application and to accept endpoints of a full scale remediation.
In-situ chemical oxidation (ISCO) by means of Fenton reagents is an established method for treating groundwater pollution. The utilization of the Fenton cycle for the production of highly reactive OH-radicals from hydrogen peroxide needs auxiliary materials. Complexing agents and/or acids are required in order to keep the catalytically active Fe- cations in the dissolved state.
In the EU-project NanoRem a variation of the conventional Fenton-based ISCO process is developed. The disadvantages of dissolved Fe are avoided by the use of Fe-loaded zeolites as catalysts. Contrary to Fe- ions as catalyst, the formation of OH-radicals at the Fe-zeolite takes place at neutral pH without the necessity for other agents. Furthermore, the zeolite framework adds sorption properties to the system, so the reactive species are formed at the centres of the highest local concentration of the target contaminant. This system offers some target selectivity due to its sorption properties towards non-polar molecules and the size-selectivity provided by the zeolite framework.
For the intended application, colloidal suspensions of Fe-zeolites with a particle size in the lower μm-range shall be injected into the subsurface where they will be immobilised on the sediment after a certain travel distance. The zeolite particles form a permeable sorption barrier which will cut off contaminant plumes by means of adsorption. After particle loading by sorption, the barrier will be regenerated by the injection of H2O2. Therefore, the injection of catalyst and oxidant is separated in time and space, offering safety and economic benefits compared to common Fenton-based ISCO. In case of plume treatment, the combination of sorption and subsequent oxidation should provide a more efficient use of H2O2 per mass of target contaminant.
Tracking Fe-zeolite particles in sediment is a very challenging task. Methods based on element composition cannot be used, since Fe-zeolites are composed of ubiquitous elements (Fe, Si and Al). Other sum parameters like turbidity only work under well-defined conditions. and iInherent zeolite properties like their high specific surface area can be used but are not applicable in a routine analysis setup.
We will present results on the issue of Fe-zeolites monitoring in water and sediment matrices. Since fluorescence as detection principle has the potential to solve many issues in tracking particle fate in the environment, a method to irreversibly label zeolites with a fluorescent dye was developed. A fluorescent dye molecule is synthesized inside the framework of the zeolites from educts which are small enough to enter the channel system. The resulting product, on the other hand, is too big to exit the channel system. This type of process is called ship-in-a-bottle synthesis. Verification of the applicability for zeolite tracing includes experiments on detection limits in various water and sediment matrices, stability against leaching and oxidants and representativeness (i.e. exclusion of impacts of labelling on particle key properties such as mobility).
Acknowledgements: This work was supported by funding from European Union within the NanoRem project.