David L. Parkhurst

Development of reaction models for ground-water systems

Abstract Methods are described for developing geochemical reaction models from the observed chemical compositions of ground water along a hydrologic flow path. The roles of thermodynamic speciation programs, mass balance calculations, and reaction-path simulations in developing and testing reaction models are contrasted. Electron transfer is included in the mass balance equations to properly account for redox reactions in ground water. The mass balance calculations determine net mass transfer models which must be checked against the thermodynamic calculations of speciation and reaction-path programs. Although reaction-path simulations of ground-water chemistry are thermodynamically valid, they must be checked against the net mass transfer defined by the mass balance calculations. An example is given testing multiple reaction hypotheses along a flow path in the Floridan aquifer where several reaction models are eliminated. Use of carbon and sulfur isotopic data with mass balance calculations indicates a net reaction of incongruent dissolution of dolomite (dolomite dissolution with calcite precipitation) driven irreversibly by gypsum dissolution, accompanied by minor sulfate reduction, ferric hydroxide dissolution, and pyrite precipitation in central Florida. Along the flow path, the aquifer appears to be open to CO 2 initially, and open to organic carbon at more distant points down gradient.

Modules based on the geochemical model PHREEQC for use in scripting and programming languages

The geochemical model PHREEQC is capable of simulating a wide range of equilibrium reactions between water and minerals, ion exchangers, surface complexes, solid solutions, and gases. It also has a general kinetic formulation that allows modeling of nonequilibrium mineral dissolution and precipitation, microbial reactions, decomposition of organic compounds, and other kinetic reactions. To facilitate use of these reaction capabilities in scripting languages and other models, PHREEQC has been implemented in modules that easily interface with other software. A Microsoft COM (component object model) has been implemented, which allows PHREEQC to be used by any software that can interface with a COM server-for example, Excel^(R), Visual Basic^(R), Python, or MATLAB^(R). PHREEQC has been converted to a C++ class, which can be included in programs written in C++. The class also has been compiled in libraries for Linux and Windows that allow PHREEQC to be called from C++, C, and Fortran. A limited set of methods implements the full reaction capabilities of PHREEQC for each module. Input methods use strings or files to define reaction calculations in exactly the same formats used by PHREEQC. Output methods provide a table of user-selected model results, such as concentrations, activities, saturation indices, and densities. The PHREEQC module can add geochemical reaction capabilities to surface-water, groundwater, and watershed transport models. It is possible to store and manipulate solution compositions and reaction information for many cells within the module. In addition, the object-oriented nature of the PHREEQC modules simplifies implementation of parallel processing for reactive-transport models. The PHREEQC COM module may be used in scripting languages to fit parameters; to plot PHREEQC results for field, laboratory, or theoretical investigations; or to develop new models that include simple or complex geochemical calculations.

Geochemical mole-balance modeling with uncertain data

Geochemical mole-balance models are sets of chemical reactions that quantitatively account for changes in the chemical and isotopic composition of water along a flow path. A revised mole-balance formulation that includes an uncertainty term for each chemical and isotopic datum is derived. The revised formulation is comprised of mole-balance equations for each element or element redox state, alkalinity, electrons, solvent water, and each isotope; a charge-balance equation and an equation that relates the uncertainty terms for pH, alkalinity, and total dissolved inorganic carbon for each aqueous solution; inequality constraints on the size of the uncertainty terms; and inequality constraints on the sign of the mole transfer of reactants. The equations and inequality constraints are solved by a modification of the simplex algorithm combined with an exhaustive search for unique combinations of aqueous solutions and reactants for which the equations and inequality constraints can be solved and the uncertainty terms minimized. Additional algorithms find only the simplest mole-balance models and determine the ranges of mixing fractions for each solution and mole transfers for each reactant that are consistent with specified limits on the uncertainty terms. The revised formulation produces simpler and more robust mole-balance models and allows the significance of mixing fractions and mole transfers to be evaluated. In an example from the central Oklahoma aquifer, inclusion of up to 5% uncertainty in the chemical data can reduce the number of reactants in mole-balance models from seven or more to as few as three, these being cation exchange, dolomite dissolution, and silica precipitation. In another example from the Madison aquifer, inclusion of the charge-balance constraint requires significant increases in the mole transfers of calcite, dolomite, and organic matter, which reduce the estimated maximum carbon 14 age of the sample by about 10,000 years, from 22,700 years to 12,600 years.

Gross-beta activity in ground water : natural sources and artifacts of sampling and laboratory analysis

Abstract Gross-beta activity has been used as an indicator of beta-emitting isotopes in water since at least the early 1950s. Originally designed for detection of radioactive releases from nuclear facilities and weapons tests, analysis of gross-beta activity is widely used in studies of naturally occurring radioactivity in ground water. Analyses of about 800 samples from 5 ground-water regions of the United States provide a basis for evaluating the utility of this measurement. The data suggest that measured gross-beta activities are due to (1) long-lived radionuclides in ground water, and (2) ingrowth of beta-emitting radionuclides during holding times between collection of samples and laboratory measurements. Although40K and228Ra appear to be the primary sources of beta activity in ground water, the sum of40K plus228Ra appears to be less than the measured gross-beta activity in most ground-water samples. The difference between the contribution from these radionuclides and gross-beta activity is most pronounced in ground water with gross-beta activities > 10 pCi/L, where these 2 radionuclides account for less than one-half the measured ross-beta activity. One exception is groundwater from the Coastal Plain of New Jersey, where40K plus228Ra generally contribute most of the gross-beta activity. In contrast,40K and228Ra generally contribute most of beta activity in ground water with gross-beta activities The gross-beta technique does not measure all beta activity in ground water. Although3H contributes beta activity to some ground water, it is driven from the sample before counting and therefore is not detected by gross-beta measurements. Beta-emitting radionuclides with half-lives shorter than a few days can decay to low values between sampling and counting. Although little is known about concentrations of most short-lived beta-emitting radionuclides in environmental ground water (water unaffected by direct releases from nuclear facilities and weapons tests), their activities are expected to be low. Ingrowth of beta-emitting radionuclides during sample holding times can contribute to gross-beta activity, particularly in ground water with gross-beta activities > 10 pCi/L. Ingrowth of beta-emitting progeny of238U, specifically234Pa and234Th, contributes much of the measured gross-beta activity in ground water from 4 of the 5 areas studied. Consequently, gross-beta activity measurements commonly overestimate the abundance of beta-emitting radionuclides actually present in ground water. Differing sample holding times before analysis lead to differing amounts of ingrowth of the two progeny. Therefore, holding times can affect observed gross-beta measurements, particularly in ground water with238U activities that are moderate to high compared with the activity of40K plus228Ra. Uncertainties associated with counting efficiencies for beta particles with different energies further complicate the interpretation of gross-beta measurements.

Parallel processing for PHAST: a three-dimensional reactive-transport simulator

A parallel algorithm for the reactive-transport simulator PHAST was developed for a Beowulf cluster of Linux PCs. The Local Area Multicomputer implementation of the Message Passing Interface standard was used for communication among processors. PHAST simulates reactive transport by operator splitting the calculation into a flow and transport step and a chemical reaction step. A load-balancing algorithm was developed using random mapping of cells to the processors for the reaction task and rebalancing after each time step. Testing with a demonstration field-scale example showed that the parallel algorithm was scalable over a range of processors and problem sizes. Load balancing is an important element of the parallel algorithm. Super-linear speedup indicated that the sequential program is not optimal in cache usage. Relative speedups on 16.8 effective processors ranged from 10 to 24 which can bring field-scale reactive-transport simulations into a reasonable timeframe.

Chemical Considerations for an Updated National Assessment of Brackish Groundwater Resources

Brackish groundwater (BGW) is increasingly used for water supplies where fresh water is scarce, but the distribution and availability of such resources have not been characterized at the national scale in the United States since the 1960s. Apart from its distribution and accessibility, BGW usability is a function of the chemical requirements of the intended use, chemical characteristics of the resource, and treatment options to make the resource compatible with the use. Here, we discuss relations between these three chemical factors using national-scale examples and local case studies. In a preliminary compilation of BGW data in the United States, five water types accounted for the major-ion composition of 70% of samples. PHREEQC calculations indicate that 57-77% of samples were oversaturated with respect to barite, calcite, or chalcedony. In the study, 5-14% of samples had concentrations of arsenic, fluoride, nitrate, or uranium that exceeded drinking-water standards. In case studies of the potential use of BGW for drinking water, irrigation, and hydraulic fracturing, PHREEQC simulations of a hypothetical treatment process resembling reverse osmosis (RO) showed that BGW had the potential to form various assemblages of mineral deposits (scale) during treatment that could adversely affect RO membranes. Speciation calculations showed that most boron in the irrigation example occurred as boric acid, which has relatively low removal efficiency by RO. Results of this preliminary study indicate that effective national or regional assessments of BGW resources should include geochemical characterizations that are guided in part by specific use and treatment requirements.

AMDTreat 5.0+ with PHREEQC titration module to compute caustic chemical quantity, effluent quality, and sludge volume

Abstract Alkaline chemicals are commonly added to discharges from coal mines to increase pH and decrease concentrations of acidity and dissolved aluminum, iron, manganese, and associated metals. The annual cost of chemical treatment depends on the type and quantities of chemicals added and sludge produced. The AMDTreat computer program, initially developed in 2003, is widely used to compute such costs on the basis of the user-specified flow rate and water quality data for the untreated AMD. Although AMDTreat can use results of empirical titration of net-acidic or net-alkaline effluent with caustic chemicals to accurately estimate costs for treatment, such empirical data are rarely available. A titration simulation module using the geochemical program PHREEQC has been incorporated with AMDTreat 5.0+ to improve the capability of AMDTreat to estimate: (1) the quantity and cost of caustic chemicals to attain a target pH, (2) the chemical composition of the treated effluent, and (3) the volume of sludge produced by the treatment. The simulated titration results for selected caustic chemicals (NaOH, CaO, Ca(OH)2, Na2CO3, or NH3) without aeration or with pre-aeration can be compared with or used in place of empirical titration data to estimate chemical quantities, treated effluent composition, sludge volume (precipitated metals plus unreacted chemical), and associated treatment costs. This paper describes the development, evaluation, and potential utilization of the PHREEQC titration module with the new AMDTreat 5.0+ computer program available at http://www.amd.osmre.gov/.ZusammenfassungGrubenwässern von Kohleminen werden häufig basische Chemikalien zudosiert mit dem Ziel der pH-Wert-Anhebung sowie zur Abtrennung von Azidität, gelöstem Aluminium, Eisen, Mangan und assoziierten Metallen. Chemikalienverbrauch und Schlammanfall beeinflussen die Betriebskosten der chemischen Wasserbehandlung. Das Computerprogramm AMDTreat, ursprünglich im Jahre 2003 entwickelt, gestattet es, derartige Kosten auf der Basis der vom Nutzer anzugebenden Menge und Beschaffenheit des unbehandelten Sauerwassers zu berechnen. Zwar kann AMDTreat die Ergebnisse empirischer Titrationstests saurer oder alkalischer Abwässer mit basischen Chemikalien verwenden, um die Behandlungskosten genau abzuschätzen, jedoch sind solche empirischen Daten nur selten verfügbar. Um die Leistungsfähigkeit von AMDTreat in Bezug auf die Abschätzung von (1) Chemikalienmenge und -kosten zur Einstellung des Ziel-pH-Werts, (2) chemischer Beschaffenheit des behandelten Wassers, und (3) Volumen der Wasserbehandlungsschlämme zu verbessern, wurde in der Version AMDTreat 5.0+ nunmehr ein Titrationssimulationsmodul auf der Basis von PHREEQC inkorporiert. Die modellierten Titrationsergebnisse für ausgewählte Alkalien (NaOH, CaO, Ca(OH)2, Na2CO3, oder NH3) können—wahlweise mit oder ohne Belüftung—mit experimentellen Daten verglichen oder aber direkt verwendet werden, um Stoffmengen, Beschaffenheit des behandelten Wassers, Schlammvolumen (gefällte Metalle plus nicht reagierte Ausgangschemikalien) sowie die sich ergebenden Behandlungskosten abzuschätzen. Der Artikel beschreibt die Entwicklung, Bewertung und potentielle Nutzung des PHREEQC-Titrationsmoduls in Verbindung mit dem neuen Computerprogramm AMDTreat 5.0+, verfügbar unter http://www.amd.osmre.gov/.ResumenSustancias alcalinas son comúnmente agregadas a descargas de minas de carbón para incrementar el pH y hacer descender la acidez y las concentraciones de aluminio, hierro, manganeso y metales asociados. El costo anual del tratamiento químico depende del tipo de sustancia y de las cantidades de sustancia usada y de lodo producido. El programa AMDTreat, desarrollado inicialmente en 2003, es ampliamente usado para calcular tales costos sobre la base a los datos de velocidad de flujo y calidad de agua especificadas para el AMD no tratado. Aunque AMDTreat puede usar resultados de la titulación empírica de la acidez o alcalinidad neta del efluente con químicos cáusticos para estimar los costos del tratamiento, tales datos empíricos están raramente disponibles. Un módulo de simulación de titulación usando el programa geoquímico PHREEQC ha sido incorporado a AMDTreat 5.0+ para mejorar la capacidad de AMDTreat para estimar: (1) la calidad y costo de químicos cáusticos para alcanzar cierto pH, (2) la composición química del efluente tratado y (3) el volumen de lodo producido por el tratamiento. Los resultados de la titulación simulada para sustancias cáusticas seleccionadas (NaOH, CaO, Ca(OH)2, Na2CO3 o NH3) sin aireación o con pre-aireación puede ser comparado con, o usada en lugar de, los datos de titulación empírica para estimar las cantidades de sustancias químicas, composición del efluente tratado, volumen del lodo (metales precipitados más sustancias químicas no reaccionantes) y costos de tratamiento asociados. Este trabajo describe el desarrollo, evaluación y potencial uso del modulo de titulación PHREEQC con el nuevo programa AMDTreat 5.0+ disponible en http://www.amd.osmre.gov/.

Phast4Windows: A 3D Graphical User Interface for the Reactive‐Transport Simulator PHAST

Phast4Windows is a Windows® program for developing and running groundwater-flow and reactive-transport models with the PHAST simulator. This graphical user interface allows definition of grid-independent spatial distributions of model properties-the porous media properties, the initial head and chemistry conditions, boundary conditions, and locations of wells, rivers, drains, and accounting zones-and other parameters necessary for a simulation. Spatial data can be defined without reference to a grid by drawing, by point-by-point definitions, or by importing files, including ArcInfo® shape and raster files. All definitions can be inspected, edited, deleted, moved, copied, and switched from hidden to visible through the data tree of the interface. Model features are visualized in the main panel of the interface, so that it is possible to zoom, pan, and rotate features in three dimensions (3D). PHAST simulates single phase, constant density, saturated groundwater flow under confined or unconfined conditions. Reactions among multiple solutes include mineral equilibria, cation exchange, surface complexation, solid solutions, and general kinetic reactions. The interface can be used to develop and run simple or complex models, and is ideal for use in the classroom, for analysis of laboratory column experiments, and for development of field-scale simulations of geochemical processes and contaminant transport.

VS2DRTI: Simulating Heat and Reactive Solute Transport in Variably Saturated Porous Media: R.W. Healy et al. Groundwater xx, no. x: xx-xx

Variably saturated groundwater flow, heat transport, and solute transport are important processes in environmental phenomena, such as the natural evolution of water chemistry of aquifers and streams, the storage of radioactive waste in a geologic repository, the contamination of water resources from acid-rock drainage, and the geologic sequestration of carbon dioxide. Up to now, our ability to simulate these processes simultaneously with fully coupled reactive transport models has been limited to complex and often difficult-to-use models. To address the need for a simple and easy-to-use model, the VS2DRTI software package has been developed for simulating water flow, heat transport, and reactive solute transport through variably saturated porous media. The underlying numerical model, VS2DRT, was created by coupling the flow and transport capabilities of the VS2DT and VS2DH models with the equilibrium and kinetic reaction capabilities of PhreeqcRM. Flow capabilities include two-dimensional, constant-density, variably saturated flow; transport capabilities include both heat and multicomponent solute transport; and the reaction capabilities are a complete implementation of geochemical reactions of PHREEQC. The graphical user interface includes a preprocessor for building simulations and a postprocessor for visual display of simulation results. To demonstrate the simulation of multiple processes, the model is applied to a hypothetical example of injection of heated waste water to an aquifer with temperature-dependent cation exchange. VS2DRTI is freely available public domain software.