Characterizing Dispersion Effectiveness at Varying Salinities

Potential oil exploration and production activities located in close proximity to brackish (low salinity) and hypersaline water bodies, such as Arctic brine channels or brine pools in the Gulf of Mexico (GoM), necessitate characterization of dispersion effectiveness (DE) of chemical dispersants with oil over a wide range of salinities. Brackish ocean waters can result in regions of ice melt. Hypersaline waters are naturally formed via evaporation, ice formation and salt deposit dissolution. Evaporation-induced hypersalinity occurs in warm shallow waters where salty, dense waters with salinities > 50 ppt are produced (e.g., Florida Bay, Everglades National Park, USA). Ice formation at high latitudes results in salt extrusion from ice matrices, where brine channels (up to 200 ppt) form at bottom of the ice (Stewart and Platford, 1986). Hypersaline waters are also found along the GoM seafloor where sediments overlie Jurassic-age Louann and Campeche salt deposits, allowing salt-tectonics to form brine pools or lakes with salinity upwards of 150 ppt (U.S. Bureau of Land Management, 1979). Pools can be tens of km long (www.oceanexplorer.gov) and found in areas with active oil and gas lease blocks. If an oil release were to occur into hypersaline waters, the extent of natural dispersion is currently unknown or whether chemical dispersants could be an effective response tool, particularly during extreme mixing of brine with overlying seawater.

Chemical dispersants contain surfactants, which reduce the oil-water interfacial tension. This promotes the formation of small oil droplets (< 100 μm) that are less likely to re-coalesce (Li and Garrett, 1998), thus suspending oil droplets as a dispersed plume. This action increases the surface area-to-volume ratio of oil droplets and enhances oil biodegradation (Venosa and Zhu, 2003; NRC, 2005). Most surfactants are more lipophilic (oil-loving) in freshwater and increase in hydrophilicity (water-loving) as the salinity rises (Li and Garrett, 1998). The stability of the resulting oil droplets is salinity dependent, increasing with ionic strength. However, above a certain salinity (surfactant dependent), this increased ionic strength results in more surfactant molecules leaving the oil droplet. Previous studies have examined this effect, whereby maximum DE was exhibited at salinities between 40 to 60 ppt (Fingas, 1991 and references therein) and was substantially lower at higher salinities above 60 ppt. The conducted experiments using DE test methods at varying dispersant-oil ratios (DOR of 1:4 through 1:25) and oil concentrations, with a range of mixing energies often insufficient for dispersion compared to the baffled flask test (BFT) (Fingas,1991; Sullivan et al., 1993; Fingas, 2004). Additionally, most studies targeted DE at low to average ocean salinity given that dispersant application would typically be considered for brackish waters (0.5 to 30 ppt), with only three studies evaluating DE at 90 ppt (Fingas, 2004).

Furthermore, despite the advantages of the BFT whose mixing is well characterized (Sorial et al., 2004; Kaku et al., 2006), the BFT does not account for the dilution of oil and dispersants that would occur at sea. Meso-scale flume and wave tanks, however, provide a remedy for this either by being sufficiently wide or by having currents/waves to simulate natural rates of dilution (Li et al., 2008; King et al., 2013; SL Ross, 2003). Recent studies indicate that DE within wave tanks is higher than within baffled flasks, regardless of how the DE is calculated or the length of time allowed for resurfacing of oil (SL Ross, 2003; Fingas, 2004). Hence, large volume tanks offer a crucial step in scaling these estimates to real world conditions. Previous wave tank studies have not included DE testing at salinities higher than average ocean salinity, as doing so requires modification to salt and water delivery systems, and the ability to properly drain and dispose of oiled brine waters. However, these modifications are possible in smaller tanks such as the Bedford Institute of Oceanography (BIO) tank facility in Dartmouth, Nova Scotia, Canada.

Decision-making regarding dispersant use during spills typically entails questions on whether there will be sufficient DE in brackish to average ocean salinity (between 0–35 ppt). Given the increased interest in leasing, exploration, and production activities in regions where hypersaline waters can be found naturally in the Arctic and Northern GoM (Macdonald et al., 1990; Joye et al., 2005), the response community is forced to ask whether dispersants could be a viable response option for subsea releases in hypersaline waters.

To address these knowledge gaps, this paper evaluates the influence of salinity on DE, droplet size, and fluorescence intensity of Alaskan North Slope (ANS) medium crude oil, with and without chemical dispersant, from 0–125 ppt salinity, using the BFT. Additionally, at larger spatial scales, ANS dispersion via droplet size was characterized during flume tank simulations of high-velocity subsea releases of oil injected with dispersant at 26, 50,75 and 100 ppt. Results help to inform spill mitigation in the Arctic and deep GoM where brine pools naturally exist.

Baffled Flask Oil Dispersion – The EPA Baffle Flask Test (BFT) procedure was used for determining DE in six 150 mL baffled trypsinizing flasks (Venosa et al., 2002; Sorial et al., 2004). Tests were conducted with and without dispersant in controlled temperature rooms (5 or 25 °C) at 0.2, 20, 35, 40, 50, 60, 75, 100 and 125 ppt salinity. Artificial seawater (120-mL) and 100 μL crude oil were added to the flask followed by 4 μL of Corexit® 9500A (Nalco, Inc.) chemical dispersant onto the slick (DOR = 1:25). Experiments were also conducted at a DOR 0 without the addition of chemical dispersant. Flasks were mixed at 250 rpm for 10 ± 0.25 minutes with settling time of 10 ± 0.25 minutes to allow undispersed oil to reform a slick prior to draining 30-mL at the flask base. The drained volume underwent liquid-liquid extraction using dichloromethane. Oil concentration was measured with a Shimadzu® UV 1800 spectrophotometer (Kyoto, Japan) between 340–400 nm. The DE (%) is calculated from the ratio of the oil dispersed to the total oil added.

The DE_LCL95 value, which is the lower 95% confidence limit of the six independent replicates, was also reported for each treatment. The remaining dispersed oil within the flask (approximately 80–85 mL) was composited (and diluted as necessary) for use in subsequent analyses for oil droplet size and in situ fluorometry, which require larger sample volumes.

Particle Size Analysis – A laser in situ scattering and transmissometry probe (LISST®-100X, Type B, Sequoia Scientific, Inc., Bellevue, Washington) was used to measure droplet sized distribution (DSD) from 1.25 to 250 μm, mean diameter (MD; also named volume mean diameter, VMD), and total volume concentration (TVC) using a small volume mixing chamber. The composited sample was stirred on a mixing plate at the test temperature until the time of measurement. Total oil droplet concentration was larger than the upper limit of detection for all oil and dispersant mixtures, thus samples were diluted at the appropriate salinity and temperature. A 100 mL sample was added to the instrument sample chamber and the built-in mixer kept the sample mixed while the measurements were made. Measurements were collected for 30 to 60 seconds and the average DSD, MD, and TVC for each treatment was summarized and reported.

Fluorescence Analysis - The composited sample was also analyzed for fluorescence intensity using a WET Labs, Inc. (Philomath, Oregon), ECO FL fluorometer wavelengths centered on excitation / emission of 370/460 nm. After 1:10 to 1:40 dilution of the composited sample, approximately 1–1.5 L of sample was used to fill a benchtop sample chamber adapter supplied by the manufacturer. Analysis of samples followed established methods for blank subtraction, and conversion to quinine sulfate dihydrate standards as outlined in the manufacturer's manual and calibration sheet. The result for each treatment was reported as ECO intensity (ppb QSDE; quinine sulfate dihydrate equivalents).

Oil Chemistry – The concentration of total petroleum hydrocarbon (TPH) of dichloromethane extracts was also measured with an Agilent (Agilent Technologies, Santa Clara, California) 7890B GC equipped with a flame ionization detector (FID) and 7693 autosampler following EPA Method 8015B. Samples were concentrated under flowing nitrogen to a final volume of 1 mL prior to analysis. The TPH in solvent extracts were determined by injecting one microliter sample in the splitless injection mode using helium as a carrier gas, with FID detector operated at 320°C and an Agilent® DB-5MS column. The concentration of the sample extracts was calculated using a six-point calibration curve generated with ANS crude oil. The average TPH for the six replicates was reported as milligrams oil dispersed.

Statistical Analysis - SigmaPlot® 14.0 was used to perform an analysis of variance to evaluate the effect of salinity on DE, fluorescence, TVC and MD.

Tank Experiments - Oil dispersion experiments were conducted using ANS oil and Corexit 9500 in the Bedford Institute of Oceanography flume tank (32 m long with operational water volume of 30,000 L). Bags of sodium chloride salt were added to the tank to create the desired salinities of 26, 50, 75 and 100 ppt. The oil injector consisted of a stainless-steel pressure vessel (2 L capacity), with oil temperature fixed at 20°C, and pressurized with compressed nitrogen to 50 psi. The nozzle orifice diameter was 2.4 mm and oriented 20 cm off the bottom of the tank and pointed in a horizontal direction to allow for using the horizontal length of the tank to capture the plume movement. Injection times were 5–7 seconds resulting in a 350 g oil release. A LISST-100X (Type C) particle size analyzer and an ECO FL fluorometer were deployed in the tank at 2.6 m downstream of the oil injection point to measure the oil plume. Expanded details of the tank and instrument suite setup can be found in Conmy et al., 2018.

Baffled Flask Oil Dispersion - Dispersion effectiveness via BFT is a measure of the total hydrocarbons dispersed into water under high mixing energy (250 rpm) and after settling time. Reported here are the DE% and DEL_CL95 values. ANS oil exposed to dispersant exhibited DE values ranging from 65.2–93.7 % in the warm temperature treatments (Table 1; Figure 1a). Oil not treated with dispersant exhibited DE values under 15 % for both temperatures. At DOR 1:25, DE decreased at salinities lower (<20 ppt) and higher (>35ppt) than average ocean salinity. Expectedly, DE was higher at 25 °C compared to 5 °C. Results suggest that the dispersant tested would be generally effective over a wide range of salinities in the presence of sufficient mixing / turbulent energy. Compared to previous studies (Fingas,1991; Sullivan et al., 1993; Fingas, 2004), the BFT also conducts at a reduced speed of 150 rpm for 35 and 100 ppt salinity treatments. The DE at 150 rpm was approximately one half (DE at 35ppt = 49.59%; DE at 100ppt = 42.84%) the value at 250 rpm. This indicates that there was less dispersion at lower mixing energy and that dispersion was hindered with increasing salinity. The TPH concentrations in the BFT treatments also support the DE results (Table 1). The DE values serve as an indication of how oil might be dispersed during a spill in the ocean or brackish water. The DE values from the BFT represent the dissolved and particulate hydrocarbon fractions of oil, which contain both aromatic (rings, multiple bonds) and aliphatic (alkanes) compounds in each fraction. The DE does not however, provide information on the relative concentrations of oil in the dissolved or droplet fractions in solution, nor does it provide for the size of the oil droplets. Oil droplet size is of value for spill preparedness as it impacts the ultimate transport and fate of spilled oil.

Droplet Size and Distribution – The droplet total volume concentration (TVC) was highest at average ocean salinities and decreased for low and high salinity treatments (Figure 1b). The mean diameter of droplets was inversely proportional to TVC (Figure 1d). The TVC as a function of droplet diameter (log transformed and ranged between 1.25 and 250 μm) for all temperature, dispersant and salinity treatments is shown in Figure 2. For oil alone treatments, droplets between 10 and 250 μm make up the biggest volume of dispersed oil, where lower temperature treatments exhibited the greatest effect. The effect of salinity was not clear for oil alone treatments. Comparatively, for the oil exposed to dispersant, salinity did impact the TVC and cumulative fraction, where increasing salinity produced a higher proportion of large droplets. At 25 °C, 90 % of the droplets were 30 μm or less for salinities above 75 ppt, whereas 90 % of droplets in the lower salinity treatments were 10 μm or less, indicating more efficient dispersion at lower salinity. An increase in droplet diameter with increasing salinity was evident for oil and dispersant treatments. Further, evaluating these results along with the BFT DE highlights that although DE for some dispersants may be above 75 % for most salinities, there was a large range in the droplet size produced. This suggests that perhaps in addition to determining the DE for an oil-dispersant combination, a measure of droplet size would inform decision-makers during a spill response.

Fluorescence – The BFT measures dispersed total hydrocarbons and the LISST measures only the particulate pool of aliphatic and aromatic compounds; whereas fluorescence measures only the aromatic compounds, both dissolved and particulate. Smaller, more volatile aromatic compounds are equated with toxicity concerns. At both temperatures, fluorescence concentration was low for the DOR 0 across the salinity range (Figure 1c). ANS with dispersant exhibited a decrease in fluorescence with increasing salinity above 20 ppt, consistent with DE.

Effect of Salinity - Previous studies have reported that the salinity of receiving waters can impact dispersion of oil by chemical dispersants (Clayton et al., 1993). Most dispersant formulations for marine use are designed to provide maximum dispersion at normal seawater salinity. The reduced DE at hypersaline conditions reported in Fingas, 2004, was observed in this study but to a lesser degree due to higher mixing speeds in the BFT (Figure 1a, Table 1). At DOR 1:25, a maximum DE of 93.7 % was observed at 20 ppt with significantly reduced DE at lower or greater salinities (p=0.018). At DOR 0, the effect of salinity on DE did not follow a clear trend (p=0.111). Similarly, fluorescence was significantly higher at 20 ppt in the presence of dispersant (p=0.003) and did not significantly vary with changes in salinity at DOR 0 (p=0.565). However, the lowest mean diameter and the highest TVC were observed at 35 ppt, followed by the values at 20 ppt. The effect of salinity on mean diameter was significant at DOR 1:25 and DOR 0 (p=0.036 and p=0.009 respectively). Overall, the effect of salinity on DE varied with DOR. While high DE values were observed at 20 ppt and DOR 1:25, there was no significant elevation at any particular salinity in the absence of a dispersant. Moreover, the effects of salinity were more apparent with “DE proxies” (fluorescence, TVC, MD) than DE measured using the traditional baffled flask test.

Flume Tank Oil Dispersion - Injection experiments were conducted with and without dispersant at 26, 50, 75 and 100 ppt salinity. Water temperature in the tank ranged between 11.7–15.3 °C. For each experiment, the oil alone injection occurred first, followed by a second injection of oil with dispersant after the initial plume was advected further downstream in the tank. During each tank treatment, continuous LISST and fluorescence data were collected to generate a time series of the plumes. Fluorescence data, calibrated to TPH and BTEX following equations in Conmy et al., 2014, exhibited higher values for DOR 1:25 compared to DOR 0, suggesting that dissolved oil concentrations in the plume were elevated due to dispersant (Table 2). The shift towards smaller droplets in the DOR 1:25 treatment compared to oil alone is shown in Figure 3 (bottom panel). Furthermore, the DSD plots for each salinity treatment (Figure 3) suggest that the presence of smaller droplets at DOR 0 was lessened at higher salinities (75 and 100 ppt). This effect was not as pronounced in the DOR 1:25 treatments. For DOR 1:25, a bimodal distribution was observed for all salinity treatments, with lowest MD of < 10 μm for the 50 ppt treatment. This suggests that oil was well dispersed over the salinity range tested. Mean diameter increased from 65 to 80 μm with increasing salinity for DOR 0, with unimodal distribution.

Results of this study suggest that ANS, a medium crude oil, was effectively dispersed with chemical dispersant over a wide range of salinities in the presence of sufficient mixing energy provided by the BFT. Similar findings would be expected for other light to medium crude oils. DE and fluorescence values were significantly higher at 20 ppt and decreased at salinities above and below this. DE does not provide information on the relative concentrations of oil in the dissolved or droplet fractions in solution, nor does it provide for the size of the oil droplets. The TVC was highest at 35 ppt, followed by 20 ppt, and decreased for low and high salinity treatments. The effect of salinity on droplet mean diameter was significant where MD increased with increasing salinity for oil and dispersant treatments, indicating less efficient dispersion. Thus, although ANS with dispersant exhibited DE values above 70 % for most salinities, there is a large range in the droplet size produced. This suggests that in addition to determining the DE for an oil-dispersant combination, a measure of droplet size would be of value for spill preparedness and for understanding the transport and fate of spilled oil.

This project was funded through an Inter-Agency Agreement No. E15PG00039 between the U.S. Department of Interior's Bureau of Safety and Environmental Enforcement (BSEE) and the U.S. Environmental Protection Agency (EPA), in collaboration with the Department of Fisheries and Oceans Canada, Bedford Institute of Oceanography, Centre for Offshore Oil, Gas and Energy Research (BIO COOGER) in Dartmouth, Nova Scotia. This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and approved for publication. Any mention of trade names, manufacturers or products does not imply an endorsement by the United States Government or the EPA. The EPA and its employees do not endorse any commercial products, services, or enterprises.