Advancing Water Resources Research and Management

ABSTRACT: Alaska fuel storage facilities are required by law to provide secondary containment for the largest tank-volume. Secondary containment at these facilities commonly includes berms, catchment basins, and ditches. Various fuel-spill scenarios influence design for secondary containment. A fuel-storage facility in Bethel, Alaska is presently undergoing secondary containment design and construction. To predict potential fuel-migration pathways, one must know the fuel-penetration rates in site soils. Field tests were conducted at the Bethel facility to measure fuel-infiltration rates in frozen soil. Laboratory tests were also conducted to measure hydraulic conductivities in frozen soils retrieved from the Bethel facility. Field and lab tests identified potential fuel-penetration depths with time.

Field tests followed ASTM Method D 5093-90 using a double-ring infiltrometer with a sealed-inner ring. Infiltration rates were measured in frozen, ice-saturated, silty sand. The average infiltration rate was 4.3x10^-8 cm/sec. Three soils were retrieved from the site and tested in the lab. A falling-head permeameter was used to measure hydraulic conductivities. Organic-rich silty sand, silty-sand fill material, and sandy silt represent the three main soil types. Samples were disturbed and compacted at densities typical of site conditions. Testing occurred at saturated and unsaturated soil-moisture contents in a cold room at -4^oC (25^oF). The permeant used for field and laboratory tests was a Diesel #2/Jet A-50 fuel mixture (heating fuel) consisting of predominantly Jet A-50. Results indicate hydraulic conductivities decrease as ice saturation increases.

KEY TERMS: hydraulic conductivity, permeability, infiltration, frozen soils, contaminant migration, secondary containment.

Secondary containment of fuel-storage facilities must protect adjacent groundwater aquifers, streams and rivers from petroleum spills. One of the largest fuel-storage facilities in the Yukon-Kuskokwim delta in western Alaska is designing their secondary containment. One aspect of the containment design relies on using frozen soil to contain fuel spills. Environmental impacts caused by a catastrophic spill on frozen soils must be evaluated. Frozen soil represents a potential matrix for use as an environmental barrier to contaminant transport. Andersland et al. (1996) studied frozen soil barriers. Their findings suggest ice-saturated soil barriers with temperatures below a contaminant's freezing point depression are effective for fuel containment.

Bethel experiences maritime weather conditions in the summer (Dorova and Hogan, 1995). They receive nearly 50% of their annual precipitation, usually in the form of drizzle, in July, August and September. Surface soil freezes near saturation in October. Freezing conditions persist near ground surface for 6 to 7 months each year. Field infiltration test and laboratory hydraulic conductivity test results help determine the effectiveness of ice-saturated soil for secondary containment of heating fuel.

Interest in contaminant transport in subsurface freezing soils is increasing. Andersland et al. (1996) report hydraulic conductivities of decane close to that of water for dry frozen soils at -10^oC (14^oF) and less than 1 X 10^-5 cm/s (3 X10^-2 ft/d) for ice-saturated soils. Wiggert et al. (1997) reported permeabilities less than 1 X 10^-9 cm² (1 X 10^-12 ft/d) for ice-saturated soils at -10^oC (14^oF) using decane as the permeant. Biggar and Neufeld (1996) found diesel migration occurred in saturated soils subjected to freeze-thaw cycles. However, diesel will not likely migrate into permanently frozen saturated soil. Biggar et al. (1998) investigated 5 sites during the summers of 1995 and 1996. Their findings indicated permafrost contamination occurred. Problematic areas were through air voids in unsaturated fill material and fissures in the ground resulting from soil contraction during freezing. Saturated permafrost exists at various depths throughout the Bethel study site.

Soil-moisture conditions influence contaminant migration in frozen soils (Anderland et al, 1996). Frozen soils consist of four main phases including air, unfrozen water, ice, and soil particles. Figure 1 illustrates the four volumetric components of frozen soil. Water volume in a soil is expressed as volumetric soil moisture content (q) using the following equation.

q = V_w/V_t

Figure 1: Volumetric distributions of frozen soil components where V_a, V_w, V_i, V_s, V_v, and V_t represent volumes of air, unfrozen water, ice, soil particles, voids, and all soil components, respectively (Modified from Andersland et al.,1996).

Tsytovich (1975) explains the soil freezing process. Soil water begins freezing when soil temperatures are below the pore water freezing point (I). Soil temperatures drop below pore waters' freezing temperatures until enough energy exists to instigate pore-water nucleation. Pore water temperature then increases to its freezing point (II). Soil temperatures remain constant at this temperature until all latent heat of fusion is released (III). Soil temperatures then decrease if ambient temperatures are below the pore-water freezing point (IV). Roman numerals represent different stages of soil freezing.

Soil-particle surface tension and capillary forces cause depressed pore water freezing temperatures adjacent to the soil grains in saturated soils. This phenomenon is most evident in fine-grain soils such as clays and silts. The unfrozen water content decreases as soil temperature decreases. The unfrozen moisture layer surrounding soil particles acts as a conduit for contaminant transport. Unsaturated frozen soils contain little or no unfrozen moisture content. Therefore, air voids in pore spaces act as conduits for contaminant transport. This is in part due to a 9% volume increase, which occurs when water freezes to form ice. Ice in the soil restricts contaminant pathways.

Fluid flow rate in porous media follows Darcy's Law (Freeze and Cherry, 1979). Darcy's Law is expressed as

Q = -KiA (2)

where Q is volumetric flow rate, K is the proportionality constant hydraulic conductivity, i is the hydraulic gradient, and A is the cross-sectional area perpendicular to flow direction. Fluid penetration rates in soils are typically expressed in velocity units as a hydraulic conductivity (K) or infiltration rate (I). Hydraulic Conductivity (K) is an empirical constant representing the rate a fluid moves through a unit cross section of saturated porous media under a unit gradient (Drainage Manual, 1978). Infiltration (I) represents the rate at which a fluid enters the soil or the fluid volume passing into the soil per unit area per unit time (Drainage Manual, 1978). The two values are identical if hydraulic gradients are the same (Freeze and Cherry, 1979). Both represent one-dimensional vertical flow.

Contaminant penetration rates in frozen soils vary depending on contaminant properties. Heating fuel, the permeant in this study, is a Light Non-Aqueous Phase Liquid (LNAPL). LNAPLs are less dense than water with negligible water solubility. They do not significantly depress water freezing point or degrade ice since water's chemical make-up is negligibly altered (Andersland et al, 1996).

LNAPL mobility in frozen soil is influenced by density and dynamic viscosity of the fluid. The greater the fluid density, the more head the contaminant will exert on underlying water, increasing hydraulic conductivities and infiltration rates. Increases in dynamic viscosity reduce hydraulic conductivities (K) and infiltration rates (I). Temperature decreases increase density and dynamic viscosities. Colder temperatures influence dynamic viscosity greater than density causing decreased LNAPL mobility in soils.

ASTM standard test method D 5093-90 was employed to determine field infiltration rates of ice-saturated soil using a double-ring infiltrometer with a sealed-inner ring. Figure 2 represents a schematic of the infiltrometer. Modifications were made on materials used and experimental dimensions to reduce soil contamination. The permeant used was a diesel #2/Jet A-50 fuel mixture instead of water. Sulfur analysis of the fuel indicated it was predominately Jet A-50. The outer-ring area was 7.53 m² (81 ft²). The inner-ring was constructed of steel rather than fiberglass to increase durability. The modified inner-ring was tightly sealed, obtaining the desired functionality. The dimensions implemented were 1.52 m² (5 ft²). Three 3000 mL capacity intravenous Vital Mix IV bags connected in parallel represented the flexible bag. The configuration allowed for fuel expansion and contraction based on fuel temperature fluctuations. Fuel infiltration into the frozen soil induced a pressure drop in the sealed-inner ring. This resulted in fuel transfer from the flexible bag configuration to the inner-ring. Equipment modifications did not alter objectives of the ASTM method.

The infiltration test location was at Bethel, Alaska's largest fuel storage tank facility near Pit #1. Volumetric soil moisture and temperature are continuously monitored in this area (Lilly and Nyman, 1999). Data collection started in April 1998. The test soil was classified as silty sand using the Unified Soil Classification System (USCS). This soil is common fill material throughout the facility. Thermistors used in pairs determined average temperatures at various locations in the infiltrometer. Two thermistors attached to the inner-ring wall measured fuel temperature. Soil temperatures were observed at the bentonite trench bottom approximately 20.3-30.5 cm (8.0-12.0 inches) below ground surface. Thermistors approximately 1.0 cm (0.4 in.) below ground surface monitored if soil thawing occurred during the experiment.

Tests performed in March and April 1999 at approximately one-week time intervals allowed for measurable fuel loss from the flexible bag configuration. Fuel expansion or contraction in the inner-ring due to temperature differences at initiation and completion of a test run was accounted for using the expansion coefficient for kerosene, 0.00049 gal/^oF (Schwendeman and Wilcox, 1987). Temperature differences between initiation and completion of all test runs were within the allowable 2.0^oC (3.6^oF) (ASTM, 1998). No detectable fuel loss was measured in the outer ring during the 40 day test. Infiltration rates (I) were calculated from the following equation:

I=167 X 10^-4(W_i-W_f)(T_f-T_i)/gtA

where I is in cm/s, W_i is the initial flexible bag mass in g, W_f is the final flexible bag mass in g, T_f and T_i represent the final and initial fuel temperatures in ^oF,g is the fuel's specific gravity, t is time duration of the test run in s, and A is the inner ring area in m². The value of 167 represents mL of fuel expansion in the inner-ring for a 0.56^oC (1.0^oF) temperature increase in the inner-ring. Masses were determined using a field balance capable of reading to the nearest gram. Fuel specific gravity was 0.84 at -2.0 ^oC (28.4^oF) using a mud-balance.

Infiltration rates were all the same order of magnitude, 10^-8 cm/s. The average infiltration rate was 4.3 X 10^-8 ±3.6 X 10^-8 cm/s (1.2 X 10^-4 ± 1.0 X 10^-4 ft/d). Figure 3 illustrates experimental results. Fuel penetration depth (x) was estimated using the relationship x = It/q. Based on the average infiltration rate, fuel penetrated an estimated 1.9 cm (0.75 in) below ground surface. This assumes an unfrozen moisture content (q) of 0.08 from Pit #1 data and 40.8 days (979.8 h) of fuel exposure time (t) to ice-saturated soil..

Hydraulic conductivity tests were performed on frozen soil samples using a falling head permeameter. Figure 4 shows the experimental set-up. Testing occurred in a cold room at -4^oC (25^oF). The permeant used was the same Diesel #2/Jet A-50 fuel mixture as used for the field infiltration test. Fuel temperature was -4^oC (25^oF). Standpipes were 50 mL burettes with a cross-sectional area (a) of 0.845 cm² (0.131 in²). Permeameter dimensions were the same as a typical compaction permeameter. Permeameter lengths (L) and areas were 11.64 cm (4.58 in) and 81.48 cm² (12.63 in²), respectively. Hydraulic conductivities (K) were computed using the following relationship:

Hydraulic conductivity tests were performed on frozen soil samples using a falling head permeameter. Figure 4 shows the experimental set-up. Testing occurred in a cold room at -4^oC (25^oF). The permeant used was the same Diesel #2/Jet A-50 fuel mixture as used for the field infiltration test. Fuel temperature was -4^oC (25^oF). Standpipes were 50 mL burettes with a cross-sectional area (a) of 0.845 cm² (0.131 in²). Permeameter dimensions were the same as a typical compaction permeameter. Permeameter lengths (L) and areas were 11.64 cm (4.58 in) and 81.48 cm² (12.63 in²), respectively. Hydraulic conductivities (K) were computed using the following relationship:

K = aLln(h₁-h₂)/A(t₂-t₁) (4)

where h₁ represents head at time t₁ and h₂ represents head at time t₂. The initial head for each test run was 100.0 cm (39.4 in). Laboratory infiltration rates in units cm/s were calculated using the following equation, a modified version of equation 3 that does not consider fuel expansion.

I = 10^-4a(h₁-h₂)/A(t₂-t₁) (5)

Variable units are cm² for a, cm for h₁ and h₂, s for t₂ and t₁, and m² for A.

Testing occurred on the 3 main soil types at the fuel storage facility. The soils, classified using the Unified Soil Classification System, were organic-rich silty sand, silty-sand fill material, and sandy silt. The silty-sand fill material is the same soil type tested during the field infiltration test. Soils were technically disturbed and dried in a convection oven at temperatures less than 15.6 ^oC (60.0^oF) to prevent oxidation of soil organic matter. Dry bulk densities (r_b) for each soil type were determined in the field. Specific gravities (G_s) were measured using the pycnometer method. The following equation shows the relationship of porosity with dry bulk density and specific gravity.

h = Void Volume/Total Volume = V_v/V_t = 1-(r_b/G_s) (6)

Moisture was added to pre-weighed quantities of dry soil and mixed thoroughly to achieve the desired volumetric moisture content (q) for unsaturated samples. Extra soil was dried in an oven between 100 and 105^oC (212 and 221^oF) to measure volumetric moisture content (q). Soil was compacted in permeameters at bulk densities determined in the field in 3 layers, each 1/3 of permeameter height. Saturated samples were achieved by first preparing an unsaturated sample. A CO₂ purge through the unsaturated soil columns followed by vacuum filtration saturated samples.

Soil samples were frozen in a cold room at -4^oC (25^oF). Measures were taken to encourage sample freezing from the top downward. Permeameters were constructed of cast acrylic. Permeameters with prepared samples were placed on a 5.1 cm (2.0 in) thick 45.7 cm (18.0 in) X 45.7 cm (18.0 in) square piece of extruded polystyrene insulation. 3 additional layers of insulation with the same dimensions were placed on top of the bottom insulation layer with holes cut in the center so the permeameters fit tightly. Only the soil surface was exposed to freezing temperatures.

Four thermistor probes inserted at 2.54 cm (1.0 in) depth increments to the center of the soil column monitored soil temperatures. Sections of the probes exposed outside of the soil column were well insulated with extruded polystyrene. Freezing curves indicate the freezing front was from the top downward.

Figure 4: Falling head permeameter testing apparatus for measuring hydraulic conductivities.

Laboratory Hydraulic Conductivity Test Results

Table 1 summarizes sample conditions and test results. Degree of ice saturation (S_i) represents the percentage of voids filled with ice. Degree of ice-saturation was calculated for each sample using the following equation.

S_i (%) = 109(V_w/V_v) = 109(q/h) (7)

The formula accounts for volume expansion of water as it freezes to form ice. Saturated samples did not contain ice-saturation over 100% because of sublimation during the freezing process. Figure 5 shows the relationship of hydraulic conductivity (K) and infiltration rate (I) with degree of ice-saturation (S_i) for different soil conditions tested. Hydraulic conductivities (K) and Infiltration rates (I) decreased with increasing degree of ice-saturation (S_i). They were close for each soil type at ice-saturation.

Figure 5: Hydraulic conductivity and infiltration rate relationship with degree of ice-saturation for Bethel soil samples.

Effectiveness of frozen soil as a barrier to a Jet A-50/Diesel #2 fuel mixture depends on degree of ice-saturation (S_i). High degrees of ice-saturation significantly reduce fuel infiltration. Although soils at low degrees of ice-saturation possess lower infiltration rates (I) than unfrozen soils, their effectiveness for restricting fuel infiltration is inadequate for secondary containment. Frozen soils with high degrees of ice-saturation significantly reduce fuel penetration. Hydraulic conductivities (K) and infiltration rates (I) are reduced to orders of magnitude of 10^-9 cm/s (10^-5 ft/d) and 10^-8 cm/s (10^-4 ft/d), respectively for ice-saturated soils. Field and Laboratory results were consistent for ice-saturated silty sand fill material. The average field infiltration rate (I) was 4.3 X 10^-8 ± 3.6 X 10^-8 cm/s (1.2 X 10^-4 ± 1.0 X 10^-4 ft/d). The average laboratory hydraulic conductivity (K) and infiltration rate (I) were 5.16E-9 ± 1.19E-9 cm/s (1.46E-5 ± 3.38E-6 ft/d) and 4.40E-8 ± 1.00E-8 cm/s (1.25E-4 ± 2.84E-5 ft/d), respectively.

Table 1: Results and soil conditions of laboratory hydraulic conductivity tests under frozen thermal regime.

The authors would like to acknowledge Bethel Fuel Sales, Inc. for financial support of this study. Fabrication of the double-ring infiltrometer with a sealed-inner ring by Mike Stanton was appreciated. We would like to thank Oly Olson and Terry Stanton for their contributions during field-testing. Assistance and advice concerning permeameter column design and machining from Eric Johansen was essential. Use of the North Pole William's Refinery sulfur analyzer by Scott Geiger and Lonney Head was appreciated.

Andersland, Orlando B., Wiggert, David C., Davies, Simon H. "Frozen Soil Subsurface Barriers: Formation and Ice Erosion." Journal of Contaminant Hydrology 23 (1996): 133-147.

Andersland, Orlando B., Wiggert, David C., Davies Simon H. "Hydraulic Conductivity of Frozen Granular Soils." Journal of Environmental Engineering March 1996: 212-216.

Biggar, Kevin W., Haidar, Saleh, Nahir, Michael, and Jarrett, Peter M. "Site Investigations of Fuel Spill Migration into Permafrost." Journal of Cold Regions Engineering June 1998: 84-104.

Biggar, K.W. and Neufeld, J.C.R. "Vertical Migration of Diesel into Silty Sand Subject to Cyclic Freeze-Thaw." Cold Regions Engineering: Proceedings of the Eighth International Conference on Cold Regions Engineering. August 12-16 1996: 116-127.

Dorava, Joseph M., and Hogan, Eppie V. "Overview of Environment and Hydrogeologic Conditions at Bethel, Alaska." U.S. Geological Survey Report 95-173 (1995)

Drainage Manual. Washington: U.S. Government Printing Office, 1978.

Freeze, R. Allan, and Cherry, John A. Groundwater. New Jersey: Prentice Hall, 1979.

Lilly, M.R., and Nyman, D.M., [1999], Environmental data for the Bethel Fuel Sales permafrost-containment project. URL:www.bethelfuels.com/environmental/data/. Anchorage, Alaska, variously paged. [May 24, 1999].

Schwendeman, Todd G. and Wilcox, H. Kendall (1987). Underground Storage Systems: Leak Detection and Monitoring. Chelsea, Michigan: Lewis Publishers, Inc.

Tsytovich, N.A. The Mechanics of Frozen Ground. Ed. George Swinzow. Washington, D.C.: Scripta Book Co. 1975.

Wiggert, David C., Andersland, Orlando B., Davies, Simon H. "Movement of Liquid Contaminants in Partially Saturated Frozen Granular Soils." Cold Regions Science and Technology 25 (1997): 111-117.

home page
Maintainer: AWRA Webserver Team
Copyright © 2000 American Water Resources Association