U.S. Geological Survey Toxic Substances Hydrology Program--Proceedings of the Technical Meeting Charleston South Carolina March 8-12, 1999--Volume 3 of 3--Subsurface Contamination From Point Sources, Water-Resources Investigations Report 99-4018C

KEYNOTE PAPER

Scale Considerations of Chlorinated Solvent Source Zones And Contaminant Fluxes: Insights From Detailed Field Studies

Chlorinated solvents are the most common contaminants of industrial origin found in groundwater in industrialized countries. They pose exceptional challenges for monitoring and remediation because they commonly penetrate deeper than other contaminants and they are persistent and mobile. The increasing dependence on risk assessment and natural attenuation for site management decisions and mass removal or in situ mass destruction for groundwater remediation has created an urgent need for improved information on the subsurface distribution of solvent mass and the contaminant mass flux in groundwater. Since 1981 the University of Waterloo has used the Borden research site situated 50 km northwest of Toronto, Ontario, for many experiments involving the release of chlorinated solvents, including experiments in which free-phase solvents (DNAPLs) are released directly into the shallow sandy aquifer. Visual examination of solvent DNAPL distributions (TCE and PCE) in excavations at Borden and in aquifer cores indicates strong influence of small-scale or subtle variations in sediment texture and at greater depth, the influence of fractures and thin discrete sandy beds in the clayey aquitard. These observations in DNAPL zones demonstrated the need for data acquisition at much smaller spatial scales than was necessary when Borden experiments involved only aqueous-phase releases for attenuation studies. The Borden site however offers only a narrow set of geological conditions, and a short time frame of contamination relative to actual contaminated sites.

In a new phase of research initiated three years ago, with continued focus on the distribution and fate of chlorinated solvents in groundwater, we have expanded the field studies to include twelve contaminated industrial sites in Canada and the United States. Also, a second site for solvent release experiments has been established on fractured clay (the Laidlaw site in southern Ontario), while use of the Borden site for sand aquifer experiments continues. Three of the industrial sites are on fractured clay, three on fractured sedimentary rock (sandstone, shale and dolomite) and the rest on surficial sand aquifers of various origins underlain by clayey aquitards. Therefore a portfolio of sites are available to us and our team of graduate students, research associates and technicians; each site offers sufficient hydrogeologic simplicity and monitoring feasibility for studies to answer specific questions in practical time frames. Except for the fractured rock sites, all of the solvent contamination occurs at depths less than 30 m in deposits easily penetrated by direct-push equipment for continuous coring and groundwater sampling. To expand the usefulness of the sites and to accomodate the variety of site conditions, existing sampling equipment is modified and new types of equipment are developed. In site selection priority is given to hydrogeologic simplicity and also their simplicity of contaminant types and input histories, which greatly enhances prospects for meaningful data interpretation. However even with this relative simplicity, the number of factors influencing the contaminant distribution and behaviour is large. The potential to accomodate more complexity in site conditions is increasing with experience and equipment improvements. The sites also must have cooperative site owners, minimal litigation activity, and good terrain access. The field studies are complemented by lab measurements of parameters such as porosity, hydraulic conductivity, effective diffusion coeffieients, fraction organic carbon and other sorption-related parameters. Large column experiments (0.3 - 0.5 m diameter) of relatively undisturbed sand or fractured clay are used to assess particular affects of geologic structure on DNAPL and solute behaviour.

At solvent DNAPL sites nearly all of the DNAPL normally resides below the water table, which causes a plume of dissolved-phase contamination to emanate from the source zone. However only in rare cases does a site offer good prospects for investigation of both the source zone and the plume. Therefore at some sites studies emphasize the source zone and at other sites the plume. Ultimately the smallest spatial scale relevant to the understanding of the behaviour and fate of solvents is the diffusion scale. This scale becomes most important for the time scales relevant to most contaminated industrial sites where DNAPL entered the subsurface decades ago. The field studies have in common the collection of thousands of samples of water and sediment or rock for chemical analyses. The samples are obtained from vetical holes to provide detailed profiles of concentration versus depth. The vertical space between samples for volatile contaminant analysis (VOC) is tailored to sediment type and age of the contamination, or in some cases sampling is done at scales as small as is feasible, which for groundwater sampling can be 0.2 - 0.4 m vertical spacing in sediment and several meters in fractured rock. Smaller vertical spacing of groundwater samples is generally not appropriate because of limitations imposed by disturbance of in situ conditions caused by purging prior to sampling. Data acquisition is accelerated by rapid on-site VOC analyses. Subsamples from sediment or rock cores are commonly taken at even smaller intervals, as close as a few millimeters in some cases. The sediment and rock core VOC analyses provide total concentrations that are connected to dissolved concentrations in pore water and concentrations sorbed on solids. In some situations the sample spacing is extremely small because the scale of diffusion effects is uncertain a priori and therefore sampling feasibility rather than prediction of contaminant distributions is the overriding factor. Networks of conventional monitoring wells exist at most of the sites and therefore comparisons are made between three sampling scales: wells, discrete-sample groundwater profiles and core subsampling.

Some preliminary conclusions can be drawn. Regardless of the depositional origin of the sandy aquifer, DNAPL accumulation zones (source zones) in sandy aquifers are comprised of many thin layers of DNAPL rather than substantial free-product pools. In some cases the DNAPL layers are associated with silt or clay layers and in other cases the control by geologic layering is much more subtle or not discernable. At locations where the DNAPL rests on top of a thick clayey layer, diffusion-scale profiles from the DNAPL into the clay are used to define the bottom of contamination and to identify advection effects and also to estimate the time of arrival of DNAPL at the location. Diffusion zones in clayey layers situated within the aquifer can comprise a large percentage of the total aqueous and sorbed mass within plumes. In sandy aquifers vertical diffusion from suspended DNAPL layers combined with lateral advection produces mixing that smooths the vertical concentration distributions within plumes short distances downgradient from the source zones. The tops and in some cases the bottoms of plumes are abrupt over vertical distances of less than a meter, which suggests diffusion control with advective influence on apparent vertical transverse dispersion.

Vertical profiles in sandy aquifers along sections (transects) across plumes at locations a short distance downgradient of the source zones show extreme variability of solvent concentrations. They show that much of the total contaminant plume mass flux occurs in small zones. Estimates of total annual contaminant mass discharge through these transects provides insight into the evolution the the source zones during previous decades and their future longevity. At sites where a network of conventional monitoring wells were used prior to our investigations, plume characteristics from these networks were generally indistinct or misleading when compared to our delineation at the finer sampling scale.

In clayey aquitards, diffusion zones (haloes) along fractures and along thin horizontal sandy layers indicate the pathways of DNAPL flow. Natural fractures in clayey deposits are commonly so small (10-50 µm apertures) that solvent DNAPL entering the fractures disappears quickly from most fractures due to diffusion. The diffusion-driven chemical mass transfer that puts the solvent mass, as aqueous and sorbed phases, into the matrix blocks between fractures. The matrix blocks, which have porosity in the 30-50% range, offer large solvent mass storage capacity relative to the fracture network. Small-scale mapping of diffusion haloes along fractures provides information on channeling of DNAPL flow and indicates variable DNAPL arrival times along fractures within the same local zone.

In fractured sedimentary rock such as sandstone the porosity of the matrix blocks (5-15%) is much smaller than the matrix porosity of the clayey deposits. However over decades this rock porosity is large enough for measurable diffusion haloes to form along the fractures where DNAPL flow or solute transport has occurred. The haloes and therefore the pathways are determined from analysis of subsamples of rock core. This approach for pathway identification has proven to be more precise and more reliable than the conventional approaches of well sampling or open borehole packer sampling that lump concentrations from various fractures, or that are influenced by borehole cross-contamination that is usually unavoidable due to natural hydraulic gradients. The rate of expansion of plumes in fractured sandstone and other sedimentary rocks can be severely restricted by the diffusion-driven chemical mass transfer from fractures where active flow occurs to the matrix blocks where the pore water is relatively immobile. Site-specific proof of this retardation of plume expansion lies in determination of the chemical mass distribution in the rock matrix.

In most of the hydrogeological environments being studied, diffusion has caused much solvent mass to enter low-permeability zones, such as clayey layers within or below sandy aquifers, or matrix blocks between fractures in clay or rock. This mass distribution causes futility of groundwater restoration efforts involving advection-based approaches such as pump-and-treat and chemical flushing. Remediation possibilities are enhanced when the technologies are capable of seeking out the solvent mass in the diffusion-controlled zones.