Spatial variation in saturated hydraulic conductivity of sediments at a crude-oil spill site near Bemidji, Minnesota

Abstract

Saturated hydraulic conductivity of aquifer sediments at a crude-oil spill research site near Bemidji, Minnesota were examined using pneumatically-induced head-difference tests and packer/vacuum system tests. Results from slug tests on 58 wells show that hydraulic conductivity varies both horizontally and vertically in the range from about 10~7 to 10~4 meters per second (m/s), with a median of 7.28 x 10~5 m/s. Hydraulic conductivities of the well-sorted medium to fine sand facies, which contains a majority of the oil plume, range from 1.76 x 10~5 to 9.82 x 10~5 m/s with a median of 5.42 x 10~5 m/s. Hydraulic conductivities of the lower sand and gravel unit, which contains a majority of the plume of dissolved petroleum constituents, range from 4.42 x 10'6 to 5.36 x 10'4 m/s with a median of 2.32 x 10'4 m/s. The average linear velocity of ground water near the spill site was calculated to examine the effects of advective flow on migration of the plumes of oil and dissolved petroleum constituents. The average linear velocity in the well-sorted medium to fine sand facies during September 1996 was about 11 meters per year (m/year). If we assume that this was the average velocity during the 17-year period since the spill (1979-96), total advective flow of ground water in this facies was about 187 m. During this 17-year period, oil floating on the water table migrated only about 40 m. By comparison, the lower sand and gravel unit had an average linear velocity of about 29 m/year, or about 3 times greater than the velocity in the well-sorted medium to fine sand facies. Based on a 29 m/year velocity, advective flow of ground water in this unit during the 17-year period since the spill was about 493 m; whereas, the plume of dissolved petroleum constituents migrated only about 200 m. These results indicate that spatial variability of hydraulic conductivity and ground-water velocity at the research site likely is a factor affecting the rate of contaminant migration. Additional research is needed to fully evaluate how the contaminant plumes are affected by changes in hydraulic properties of the various lithologic units. Introduction Knowledge of the hydraulic properties of sediments is important in evaluating the distribution and movement of nonaqueous phase liquids (NAPL), such as petroleum, through an aquifer. Variations in hydraulic conductivity within an aquifer likely affect the extent and migration of the NAPL plume, as well as the plume of dissolved petroleum constituents. Information on the effects of spatial variations in hydraulic properties of sediments on the distribution of NAPL have been collected at several study sites (Abdul and others, 1990; Ashley and others, 1994; Johnston and Patterson, 1994). The effects of spatial variations in hydraulic properties of sediments on the movement of NAPL and water within aquifers have been studied by Faust and others (1989), Kueper and Frind (1991), and Kueper and others (1993). On August 20, 1979 approximately 16 km northwest of Bemidji, Minnesota, the land surface and shallow subsurface were contaminated when a crude-oil pipeline burst, spilling 1,700,000 L of crude oil onto a glacial outwash aquifer (fig. 1). After cleanup efforts were completed, 400,000 L of crude oil remained in the ground (Hull, 1984). Some crude oil percolated through the unsaturated zone to the water table near the pipeline break, forming the North oil pool. Crude oil also sprayed to the southwest of the pipeline break and moved over topographically lower areas to form a second large area of oil contamination on the water table, the South oil pool. This crude-oil spill has been the location of hydrological, biological and chemical studies examining the movement and fate of NAPL in the ground-water system since 1983 (Hult, 1984). Research at the North oil pool has focused on characterization of the hydrogeochemical environment, and on identification of the controlling geochemical reactions (Baedecker and others, 1993). Crude oil, water, soil, vapor, and biological samples have been collected and analyzed. Many of the studies, such as those utilizing solute transport models, models of the chemical evolution within the oil plume, and determination of field biodegradation rates, are dependent on accurate determination of the spatial variations in hydraulic conductivity of the sediments. Geochemical reactions in the surficial sand and gravel aquifer that affect the crude oil are controlled by dissolution, degradation, and transport of organic compounds (Baedecker and others, 1993). The distribution of unstable constituents that are reactants in, or products of, reactions with organic compounds, have been used to delineate five zones within the aquifer where specific geochemical processes occur (Baedecker and others, 1993). The five zones are as follows: (1) oxygenated, uncontaminated native ground water; (2) reduced oxygen ground water beneath the spray area; (3) anoxic ground water containing high concentrations of hydrocarbons, dissolved manganese and iron, and methane beneath and immediately downgradient from the oil lens; (4) a transition zone between oxygenated and anoxic ground water where hydrocarbon concentrations decrease rapidly; and (5) oxygenated ground water downgradient of the contamination plume containing slightly elevated concentrations of dissolved constituents. Information on the hydraulic properties of sediments in these five zones is important for geochemical modeling of processes affecting the crude oil distribution and degradation. Attempts to determine the hydraulic properties of aquifer sediments at the research site have been limited by the need to minimize the effects of testing on the (1) water table, (2) contaminant plume, and (3) geochemical zones (Baedecker and others, 1993). Therefore, large-scale aquifer tests using a highcapacity well that may interfere with natural ground-water flow within the aquifer have been avoided. Two smallerscale aquifer tests were conducted at the site during the 1980's and encountered difficulties due to inaccuracies in estimating aquifer thickness and to the large degree of heterogeneity within the aquifer (H.I. Essaid, U.S. Geological Survey, written commun., 1997). Therefore, investigations have largely centered upon simpler approaches, such as calculation of saturated hydraulic conductivities based on measured groundwater velocities and grain-size analyses (White, 1991), and based on permeameter tests and grain-size analyses (Bilir, 1992). Fluid saturation, particle-size distribution, and porosity measurements have been used to examine the effect of permeability distribution on multiphase (NAPL, water, and air) flow (Dillard and others, 1997). The effects of spatial variability of hydraulic properties on fluid distributions at the site have also been simulated by Essaid and others (1993). This report describes the results of slug tests used to determine the horizontal and vertical distribution of saturated hydraulic conductivity of sediments at the research site. The effects of varying ground-water velocities in the various geologic units is also discussed in relation to the plumes of oil and dissolved petroleum constituents. Slug tests provide a means of measuring saturated hydraulic conductivities while causing only minor effects on the configuration of the water table. Slug tests are an appropriate method for examining the hydraulic properties of the aquifer because the thickness and lithology of the sediments are known at each test site, and slug tests have a limited radial influence on heads in the aquifer around each well. The research site is located on a stratified, pitted and dissected glacial outwash aquifer. In general, the aquifer is 7-23 m thick and is composed mainly of moderately well-sorted to poorly-sorted sand. The aquifer is interbedded with discontinuous bodies of diamicton, sediment-flow deposits, and lacustrine silt and clay (Franzi, 1987). The aquifer is underlain by a diamicton layer at depths of 23-31 m. The upper unit of the aquifer along cross section A-A' near the North oil pool (fig. 2) is an approximately 2.5-8 m thick, poorly sorted, interbedded coarse sand, gravel, and silt facies. Underlying this facies is an approximately 1-10 m thick, well-sorted medium to fine sand facies that contains 0-5 m thick lenses of interbedded, poorly-sorted coarse sand and gravel. A discontinuous 0-3 m thick unit of laminated silt, very fine sand, and clay underlies the well-sorted medium to fine sand facies in the northeastern part of the cross section, and in the southwestern part of the cross section between wells E X P L A N A T IO N W et la nd K no w n ap pr ox im at e ex te nt of o il po ol , S ep te m be r 19 96 A ' Li ne o f s ec tio n, o bs er va tio n w el ls i nc lu de d in s ec tio n ar e in b ol d ty pe .