Colin D. Johnston

Aerobic bioremediation of 1,2 dichloroethane and vinyl chloride at field scale

Aerobic bioremediation of 1,2 dichloroethane (1,2 DCA) and vinyl chloride (VC) was evaluated at field scale in a layered, silty and fine-sand anaerobic aquifer. Maximum concentrations of 1,2 DCA (2 g/L) and VC (0.75 g/L) in groundwater were within 25% and 70% of pure compound solubility, respectively. Aerobic conditions were induced by injecting air into sparging wells screened 20.5-21.5 m below ground (17-18 m below the water table). Using a cycle of 23 h of air injection followed by three days of no air injection, fifty days of air injection were accumulated over a 12 month period which included some longer periods of operational shutdown. Oxygen and volatile organic compound probes, and multilevel samplers were used to determine changes of the primary contaminants and the associated inorganic chemistry at multiple locations and depths. Air (oxygen) was distributed laterally up to 25 m from the sparge points, with oxygen partial pressures up to 0.7 atmospheres (28-35 mg/L in groundwater) near to the sparge points. The dissolved mass of 1,2 DCA and VC was reduced by greater than 99% over the 590 m(2) trial plot. Significantly, pH declined from nearly 11 to less than 9, and sulfate concentrations increased dramatically, suggesting the occurrence of mineral sulfide (e.g., pyrite) oxidation. Chloride and bicarbonate (aerobic biodegradation by-products) concentration increases were used to estimate that 300-1000 kg of chlorinated hydrocarbons were biodegraded, although the ratio of 1,2 DCA to VC that was biodegraded remained uncertain. The mass biodegraded was comparable but less than the 400-1400 kg of chlorinated compounds removed from the aqueous phase within a 10,000 m(3) volume of the aquifer. Due to the likely presence of non-aqueous phase liquid, the relative proportion of volatilisation compared to biodegradation could not be determined. The aerobic biodegradation rates were greater than those previously estimated from laboratory-based studies.

Dissolution of multi-component LNAPL gasolines: The effects of weathering and composition

The composition of light non-aqueous phase liquid (LNAPL) gasoline and other petroleum products changes profoundly over their life once released into aquifers. However limited attention has been given to how such changes affect key parameters such as the activity coefficients which control partitioning of components of petroleum fuel into groundwater and are used to predict long-term risk from fuel releases. Laboratory experiments were conducted on a range of fresh, weathered and synthetic gasoline mixtures designed to mimic the expected changes in composition in an aquifer. Weathered gasoline created under controlled evaporation and water washing, and naturally weathered gasoline, were investigated. Equilibrium concentrations in water and molar fractions in the gasoline mixtures were compared with equilibrium concentrations predicted by Raoults law assuming ideal behaviour of the solutions. The experiments carried out allowed the relative sensitivity of the activity coefficients of key risk drivers such as benzene, toluene, ethylbenzene and xylene (BTEX) compounds to be quantified with respect to the presence of other types of compounds and where the source LNAPL had undergone different types of weathering. Results differed for the mixtures examined but in some cases higher than predicted dissolved equilibrium concentrations showed non-ideal behaviour for toluene, benzene and xylenes. Comparison of the activity coefficients showed that the naturally weathered gasoline and a 50% evaporated unleaded gasoline present a similar range of values varying between 1.0 and 1.2, suggesting close to ideal partitioning between the LNAPL and water. The fresh and water-washed gasoline had higher values for the activity coefficient, from 1.2 to 1.4, indicating non-ideal partitioning. Results from synthetic mixtures demonstrated that these differences could be due to the different molar fractions of the nC5 and nC6 aliphatic hydrocarbons acting on the molecular interactions, while differences in molar volumes seemed to have less of an influence on ideality.

The use of mass depletion-mass flux reduction relationships during pumping to determine source zone mass of a reactive brominated-solvent DNAPL

Mass depletion-mass flux relationships usually applied to a groundwater plume were established at field scale for groundwater pumped from within the source zone of a dense non-aqueous phase liquid (DNAPL). These were used as part of multiple lines of evidence in establishing the DNAPL source mass and architecture. Simplified source mass-dissolved concentration models including those described by exponential, power, and error functions as well as a rational mass equation based on the equilibrium stream tube approach were fitted to data from 285 days of source zone pumping (SZP) from a single well which removed 152 kg of dissolved organics from a multi-component, reactive brominated solvent DNAPL. The total molar concentration of the source compound, tetrabromoethane and its daughter products was used as a single measure of contaminant concentration to relate to source mass. A partitioning inter-well tracer test (PITT) conducted prior to the SZP provided estimates of groundwater travel times, enabling parameterisation of the models. After accounting for capture of the down-gradient dissolved plume, all models provided a good fit to the observed data. It was shown that differentiation between models would only emerge after appreciably more pumping from the source zone. The model fits were not particularly sensitive to the exponent parameters and variance of groundwater travel time. In addition, the multi-component nature of the DNAPL did not seem to affect the utility of the models for the period examined. Estimates of the DNAPL mass prior to the start of SZP from the models were greatest where the log of the variance of travel time was used explicitly in the source depletion models (mean 295kg) compared to where the associated power exponent and variance was fitted freely (mean 258 kg). The estimates of source mass were close to that of 220kg determined from the PITT. In addition to the PITT, multi-level groundwater sampling from within the source zone provided important supporting information for developing the conceptual model of the source zone. It is concluded that SZP may be an effective and relatively simple means for characterising DNAPL source zones.

Integrating spatial and temporal oxygen data to improve the quantification of in situ petroleum biodegradation rates

Accurate estimation of biodegradation rates during remediation of petroleum impacted soil and groundwater is critical to avoid excessive costs and to ensure remedial effectiveness. Oxygen depth profiles or oxygen consumption over time are often used separately to estimate the magnitude and timeframe for biodegradation of petroleum hydrocarbons in soil and subsurface environments. Each method has limitations. Here we integrate spatial and temporal oxygen concentration data from a field experiment to develop better estimates and more reliably quantify biodegradation rates. During a nine-month bioremediation trial, 84 sets of respiration rate data (where aeration was halted and oxygen consumption was measured over time) were collected from in situ oxygen sensors at multiple locations and depths across a diesel non-aqueous phase liquid (NAPL) contaminated subsurface. Additionally, detailed vertical soil moisture (air-filled porosity) and NAPL content profiles were determined. The spatial and temporal oxygen concentration (respiration) data were modeled assuming one-dimensional diffusion of oxygen through the soil profile which was open to the atmosphere. Point and vertically averaged biodegradation rates were determined, and compared to modeled data from a previous field trial. Point estimates of biodegradation rates assuming no diffusion ranged up to 58 mg kg(-1) day(-1) while rates accounting for diffusion ranged up to 87 mg kg(-1) day(-1). Typically, accounting for diffusion increased point biodegradation rate estimates by 15-75% and vertically averaged rates by 60-80% depending on the averaging method adopted. Importantly, ignoring diffusion led to overestimation of biodegradation rates where the location of measurement was outside the zone of NAPL contamination. Over or underestimation of biodegradation rate estimates leads to cost implications for successful remediation of petroleum impacted sites.

Evaluating the reliability of equilibrium dissolution assumption from residual gasoline in contact with water saturated sands

Understanding dissolution dynamics of hazardous compounds from complex gasoline mixtures is a key to long-term predictions of groundwater risks. The aim of this study was to investigate if the local equilibrium assumption for BTEX and TMBs (trimethylbenzenes) dissolution was valid under variable saturation in two dimensional flow conditions and evaluate the impact of local heterogeneities when equilibrium is verified at the scale of investigation. An initial residual gasoline saturation was established over the upper two-thirds of a water saturated sand pack. A constant horizontal pore velocity was maintained and water samples were recovered across 38 sampling ports over 141days. Inside the residual NAPL zone, BTEX and TMBs dissolution curves were in agreement with the TMVOC model based on the local equilibrium assumption. Results compared to previous numerical studies suggest the presence of small scale dissolution fingering created perpendicular to the horizontal dissolution front, mainly triggered by heterogeneities in the medium structure and the local NAPL residual saturation. In the transition zone, TMVOC was able to represent a range of behaviours exhibited by the data, confirming equilibrium or near-equilibrium dissolution at the scale of investigation. The model locally showed discrepancies with the most soluble compounds, i.e. benzene and toluene, due to local heterogeneities exhibiting that at lower scale flow bypassing and channelling may have occurred. In these conditions mass transfer rates were still high enough to fall under the equilibrium assumption in TMVOC at the scale of investigation. Comparisons with other models involving upscaled mass transfer rates demonstrated that such approximations with TMVOC could lead to overestimate BTEX dissolution rates and underestimate the total remediation time.

Mass discharge assessment at a brominated DNAPL site: Effects of known DNAPL source mass removal

Management and closure of contaminated sites is increasingly being proposed on the basis of mass flux of dissolved contaminants in groundwater. Better understanding of the links between source mass removal and contaminant mass fluxes in groundwater would allow greater acceptance of this metric in dealing with contaminated sites. Our objectives here were to show how measurements of the distribution of contaminant mass flux and the overall mass discharge emanating from the source under undisturbed groundwater conditions could be related to the processes and extent of source mass depletion. In addition, these estimates of mass discharge were sought in the application of agreed remediation targets set in terms of pumped groundwater quality from offsite wells. Results are reported from field studies conducted over a 5-year period at a brominated DNAPL (tetrabromoethane, TBA; and tribromoethene, TriBE) site located in suburban Perth, Western Australia. Groundwater fluxes (qw; L(3)/L(2)/T) and mass fluxes (Jc; M/L(2)/T) of dissolved brominated compounds were simultaneously estimated by deploying Passive Flux Meters (PFMs) in wells in a heterogeneous layered aquifer. PFMs were deployed in control plane (CP) wells immediately down-gradient of the source zone, before (2006) and after (2011) 69-85% of the source mass was removed, mainly by groundwater pumping from the source zone. The high-resolution (26-cm depth interval) measures of qw and Jc along the source CP allowed investigation of the DNAPL source-zone architecture and impacts of source mass removal. Comparable estimates of total mass discharge (MD; M/T) across the source zone CP reduced from 104gday(-1) to 24-31gday(-1) (70-77% reductions). Importantly, this mass discharge reduction was consistent with the estimated proportion of source mass remaining at the site (15-31%). That is, a linear relationship between mass discharge and source mass is suggested. The spatial detail of groundwater and mass flux distributions also provided further evidence of the source zone architecture and DNAPL mass depletion processes. This was especially apparent in different mass-depletion rates from distinct parts of the CP. High mass fluxes and groundwater fluxes located near the base of the aquifer dominated in terms of the dissolved mass flux in the profile, although not in terms of concentrations. Reductions observed in Jc and MD were used to better target future remedial efforts. Integration of the observations from the PFM deployments and the source mass depletion provided a basis for establishing flux-based management criteria for the site.

Effect of compositional heterogeneity on dissolution of non-ideal LNAPL mixtures.

The extent of dissolution of petroleum hydrocarbon fuels into groundwater depends greatly on fuel composition. Petroleum fuels can consist of thousands of compounds creating different interactions within the non-aqueous phase liquid (NAPL), thereby affecting the relative dissolution of the components and hence a groundwater plumes composition over long periods. Laboratory experiments were conducted to study the variability in the effective solubilities and activity coefficients for common constituents of gasoline fuels (benzene, toluene, p-xylene and 1,2,4-trimethylbenzene) (BTX) in matrices with an extreme range of molar volumes and chemical affinities. Four synthetic mixtures were investigated comprising BTX with the bulk of the NAPL mixtures made up of either, ethylbenzene (an aromatic like BTX with similar molar volume); 1,3,5-trimethylbenzene (an aromatic with a greater molar volume); n-hexane (an aliphatic with a low molar volume); and n-decane (an aliphatic with a high molar volume). Equilibrium solubility values for the constituents were under-predicted by Raoults law by up to 30% (higher experimental concentrations) for the mixture with n-hexane as a filler and over-predicted by up to 12% (lower experimental concentrations) for the aromatic mixtures with ethylbenzene and 1,3,5-trimethylbenzene as fillers. Application of PP-LFER (poly-parameter linear free energy relationship) model for non-ideal mixtures also resulted in poor correlation between experimentally measured and predicted concentrations, indicating that differences in chemical affinities can be the major cause of deviation from ideal behavior. Synthetic mixtures were compared with the dissolution behavior of fresh and naturally weathered unleaded gasoline. The presence of lighter aliphatic components in the gasoline had a profound effect on estimating effective solubility due to chemical affinity differences (estimated at 0.0055 per percentage increase in the molar proportion of aliphatic) as well as reduced molar volumes (estimated at -0.0091 in the activity coefficient per unit increase in molar volume, mL/mol). Previously measured changes in activity coefficients due to natural weathering of 0.25 compares well to 0.27 calculated here based on changes in the chemical affinity and molar volumes. The study suggests that the initial estimation of the composition of a fuel is crucial in evaluating dissolution processes due to ideal and non-ideal dissolution, and in predicting long term dissolution trends and the longevity of NAPL petroleum plume risks.

Impact of Rock Heterogeneity on Interactions of Microbial-Enhanced Oil Recovery Processes

Residual oil saturation reduction and microbial plugging are two crucial factors in microbial-enhanced oil recovery (MEOR) processes. In our previous study, the residual saturation was defined as a nonlinear function of the trapping number, and an explicit relation between the residual oil saturation and the trapping number was incorporated into a fully coupled biological (B) and hydrological (H) finite element model. In this study, the BH model is extended to consider the impact of rock heterogeneity on microbial-enhanced oil recovery phenomena. Numerical simulations of core flooding experiments are performed to demonstrate the influences of different parameters controlling the onset of oil mobilization. X-ray CT core scans are used to construct numerical porosity-permeability distributions for input to the simulations. Results show clear fine-scale fingering processing, and that trapping phenomena have significant effects on residual oil saturation and oil recovery in heterogeneous porous media. Water contents and bacterial distributions for heterogeneous porous media are compared with those for homogenous porous media. The evolution of the trapping number distribution is directly simulated and visualized. It is shown that the oil recovery efficiency of EOR/MEOR will be lower in heterogeneous media than in homogeneous media, largely due to the difficulty in supplying surfactant to unswept low-permeability zones. However, MEOR also provides efficient plugging along high-permeability zones which acts to increase sweep efficiency in heterogeneous media. Thus, MEOR may potentially be more suited for highly heterogeneous media than conventional EOR.