Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 135
Oil Spill Dispersants: Efficacy and Effects 4 Transport and Fate Spilled oil is transported, and its composition and character altered, by a variety of physical, chemical, and biological processes (Figure 4-1). Use of chemical dispersants changes the relative importance of these processes, affecting the fate of the oil, and altering subsequent ecological effects. Thus, it is important to understand the transport and fate of oil with and without dispersant use. A number of comprehensive studies have reviewed these mechanisms including Stolzenbach et al. (1977), Kerr and Barrientos (1979), Huang and Monastero (1982), Payne and McNabb (1984), Payne et al. (1984), Delvigne et al. (1986), Spaulding (1988), Lee et al. (1990), Payne et al. (1991a,b,c,d), Yapa and Shen (1994), ASCE (1996), Reed et al. (1999), Payne and French-McCay (2001), Payne and Driskell (2003), and NRC (1985, 1989, 2003). These mechanisms are reviewed briefly in the first two sections of this chapter, with a focus on how transport and fate influence the subsurface concentration of oil, and how the composition and concentration of surface and entrained oil droplets can be expected to vary with and without application of chemical dispersants. The latter portion of the chapter and Appendix E discuss how the mechanisms are integrated into computer oil spill models that simulate the fate of spilled oil, and how such models are used (or might be used) for purposes of pre-planning, emergency response, and natural resource damage assessment. TRANSPORT PROCESSES There are three major modes of transport for spilled oil or petroleum products discussed in the following subsections. The first deals with the
OCR for page 136
Oil Spill Dispersants: Efficacy and Effects FIGURE 4-1 Major open-ocean oil fate and transport processes. SOURCE: NRC, 1985. surface transport of slicks, which is important because the shape, thickness, and location of a slick affect the ability to effectively apply dispersants. The second subsection deals with vertical transport, which is responsible for the initial dilution of dispersed oil. Finally, the last subsection deals with horizontal subsurface transport, which is responsible for the ultimate dilution of dispersed oil. Surface Transport Oil spilled directly on a calm water surface spreads radially by gravity and is resisted by inertia, viscosity, and surface tension until the slick reaches a thickness of ~0.1 mm. Fay (1969), Hoult (1972), and others have modeled this spreading under idealized conditions (e.g., instantaneous spill, no wind, no waves). Application of chemical dispersants can temporarily affect this spreading through the phenomenon of herding.
OCR for page 137
Oil Spill Dispersants: Efficacy and Effects Additional spreading takes place because (1) oil is usually spilled over a period of time and into a moving current, (2) wind, waves, and non-uniform currents diffuse and break up the slick, and (3) droplets periodi-cally disperse and resurface, tending to stretch the plume. This last mechanism has been described by Johansen (1984) and Elliott et al. (1986) and may increase in significance when considering the fate of chemically dispersed oil. Slick thicknesses were estimated during several well-documented oil spills, usually indirectly by dividing volume/area (Mackay and Chau, 1986; Lunel and Lewis, 1993a,b; Lewis et al., 1995a,b; Walker et al., 1995; Brandvik et al., 1996; Brown et al., 2000). These studies indicate that oil does not spread uniformly, but is irregular in shape and thickness—generally elongated in the direction of the wind and often composed of thick patches (>1 mm) and thinner sheens (<0.01 mm). S.L. Ross (1997) gives a general rule of thumb that 90 percent of an oil spill’s volume is contained in 10 percent of its area. Figure 2-5 (in Chapter 2) presents representative descriptions of the wide range of slick thicknesses typical of an oil slick along with an approximation of the estimated volume/unit area for the different thicknesses. The non-uniform characteristics of a slick can be included in models (e.g., Mackay et al., 1980a,b; Lehr et al., 1984), but such models are basically empirical. Surface spreading has important implications for the operational effectiveness of dispersant application because dispersant delivery systems have finite encounter rates (area coated per unit time) and capacities (total volume of dispersant used; Gregory et al., 1999). As such, dispersants are most effective when they are applied as soon as possible (before the slick has had time to spread and break up), and with the benefit of airborne sensing to identify locations of maximum slick thickness. In particularly thick regions, it is not practical to treat the slicks with a single pass and lacking visual confirmation of dispersion, a multi-pass approach is often used (S.L. Ross, 1997; Lunel et al., 1997b). Of additional concern is oil that is accidentally released from subsurface blowouts during offshore exploration or production. Here the oil will likely be mixed with substantial quantities of natural gas, which provides the major source of buoyancy. Masutani and Adams (2004) and Tang (2004) describe the spectrum of oil droplet sizes that can be expected as a function of dimensionless exit conditions. The combination of gas and oil forms a buoyant droplet/bubble plume that entrains seawater as it ascends toward the surface. A similar situation, but without the gas, would occur with the rupture of an underwater oil pipeline. Models to describe such plumes have been developed by Yapa and Zheng (1997; 1999) and Johansen (2000) among others. If the oil is released in shallow water (less than roughly 100 m), it will rise as a coherent plume, containing a mixture
OCR for page 138
Oil Spill Dispersants: Efficacy and Effects of gas, oil, and water. Once the plume surfaces, the oil and water will spread radially in a surface layer (Fannelop and Sjoen, 1980). Because of the presence of water, the resulting oil slick will be significantly thinner than those produced by oil spilled directly on the surface. In deeper waters, ambient currents, and potentially density stratification, will cause the gas bubbles and larger oil droplets to separate from the remainder of the plume and ascend as individual (or small groups of) droplets and bubbles (Socolofsky and Adams, 2002). Because droplet rise velocity depends on diameter, the larger oil droplets will reach the surface sooner and closer to their source than the smaller droplets. This fractionation leads to a substantially longer (and thinner) plume than would be produced by a surface spill. Work is being conducted both in the United States and abroad, to assess if and how to chemically disperse oil from a subsurface blowout. In many cases, it is impractical to apply dispersants at the surface because the slick is too thin. However, if the surface slick is subsequently concentrated by Langmuir circulation cells or other convergence mechanisms, dispersants can be applied to the thicker portions. In the absence of such surface convergence, the most effective method would be to apply dispersant within the well (down hole) before the oil can mix with seawater, but this may be difficult, so attention is being paid to schemes that dispense the dispersants directly into the plume. This should be done as close to the seafloor as possible to minimize dilution, and hence achieve the desired dispersant-to-oil ratio (DOR) without bearing the cost and potential environmental consequences of using excessive quantities of dispersant. Some initial concepts for dispersant application to blowouts can be found in Johansen and Carlsen (2002). Slicks are advected downwind by a combination of wind and waves. Pure advection (without spreading) does not affect the concentration of oil or the effectiveness of dispersants, but it is important for understanding where an oil slick will end up. Many researchers have studied these processes from theoretical and empirical perspectives, and a rule of thumb is that slicks move at approximately 3 percent of the wind speed measured 10 m above the water surface (i.e., the “wind factor” is about 3 percent). For moderate to high sea states, approximately two-thirds of this transport can be attributed to Stokes drift (the fact that near-surface wave orbits in deep water waves do not follow exact circles, as linear theory would suggest, but exhibit a net transport in the direction of wave propagation). The remaining one-third represents the slick moving relative to the water directly underneath it (Lehr et al., 2002). Coriolis acceleration causes the slick to drift ~10–20 percent to the right of the wind in the northern hemisphere, but this effect is often omitted in transport models. Experimental observations support these conclusions, with some sugges-
OCR for page 139
Oil Spill Dispersants: Efficacy and Effects tion that the wind factor and deflection angle decline with wind speed (Youssef and Spaulding, 1993). Vertical Transport Dispersion of a surface slick, whether caused naturally or through application of chemical dispersants, results in the formation of droplets that are entrained into the water column and transported with the subsurface currents. The importance of vertical transport is clearly seen by a simple calculation for illustrative purposes: a surface slick that is 0.1 mm thick and dispersed with an efficiency of 50 percent to an average depth of 5 m, will receive a dilution of 105, resulting in an immediate drop in concentration to ~10 ppm. Dispersion results in a distribution of droplet sizes with the smaller droplets being transported deeper and longer. If Q is the mass of oil entrained per unit area of the slick, and d is a characteristic droplet diameter, it is clear that the goal of chemical dispersants is to increase Q and decrease d. And while it is obvious that use of chemical dispersants increases the mass of oil within the water column, it may or may not increase the concentration of oil, because the greater dilution achieved by smaller droplets may offset the increase in mass. This question will be revisited at the end of this subsection. The initial depth of droplet penetration, hi, is proportional to the wave height, hw, with many studies showing that (Nilsen et al., 1985; Delvigne and Sweeney, 1988). [Variables used in this chapter are summarized in Table 4-1.] Subsequent vertical transport depends on a balance between vertical diffusion (characterized by a vertical diffusivity Ez, with dimensions of L2/T) and buoyant rise (characterized by a terminal velocity ws). Vertical diffusivity transports droplets deeper into the water column, while buoyancy makes them return to the surface. Vertical diffusivity generally ranges between 1 and 200 cm2/s depending on a number of environmental factors. Near the surface, diffusivity is a strong function of wave height, and a number of investigators report Ez ~ hw2 (Koh and Fan, 1970). Because wave energy decreases with depth, Ez decreases below the surface. For example, Ichiye (1967) suggests that, in the absence of density stratification, (4-1) where L is wave length, T is wave period, and hw is taken as the significant wave height. Other formulations suggest a stronger cut-off with depth, attributed to the depth of Langmuir circulation (windrows), which is caused by the interaction of wind and waves (Leibovich and Lumley, 1982;
OCR for page 140
Oil Spill Dispersants: Efficacy and Effects TABLE 4-1 Variables Used in Scaling Arguments in Chapter 4 Variable Definition Dimension Cdiss Oil concentration in dissolved phase ML−3 Cdrop Oil concentration in droplet phase ML−3 d Droplet diameter L Ez Vertical diffusion coefficient L2T−1 Er Horizontal (radial) diffusion coefficient L2T−1 hchar Characteristic depth of oil droplets L hi Initial depth of oil droplets L hw Wave height L L Wave length L Q Mass of oil entrained per unit area of slick ML−2 T Wave period T ws Droplet slip (rise) velocity LT−1 λz Vertical velocity gradient T−1 σr Radial standard deviation of spreading patch L ν Kinematic viscosity L2T−1 P1 Water density ML−3 and references in Champ, 2000). Diffusivity also decreases under the influence of vertical density stratification, and a host of formulations suggest that Ez is inversely proportional to the vertical density gradient (Koh and Fan, 1970; Broecker and Peng, 1982). A thermocline is a region of maximum density gradient suggesting small Ez, and if stratification is strong enough, a “diffusion floor” may be assumed. Some models assume that the depth of this floor is simply proportional to wave height. Unless there is significant interaction with suspended particulates, most oil droplets will be positively buoyant and will rise toward the surface. Those with a diameter less than about 300 mm will obey Stokes Law and rise with a velocity of: (4-2) where ν is the kinematic viscosity of water, Δρ/ρ is the normalized density difference between seawater and oil, g is gravitational acceleration, and d is droplet diameter. The quadratic dependence of rise velocity on droplet diameter suggests that the smallest droplets will rise very slowly, accentuating dispersion. For example, with Δρ/ρ = 0.13 (for an oil with a density of 0.89 mg/mL and seawater at 1.025 mg/mL), ν = 10−2 cm2/s and g = 981 cm/s2, droplets with a diameter of 300µm will rise with a velocity of 0.6 cm/s while droplets with a diameter of 30 µm will rise with a velocity of 0.006 cm/s. The former will take less than 8 minutes to rise a height
OCR for page 141
Oil Spill Dispersants: Efficacy and Effects of 3 m, while the latter will take over 12 hours. And, because of vertical diffusion, the smaller droplets will most likely reside deeper in the water column, further prolonging their ascent. The above discussion can be used to estimate how the concentration of droplet and dissolved phase oil might depend on dispersion efficiency and vertical transport mechanisms. The concentration of oil in the droplet phase is proportional to the mass of oil entrained per unit area, Q, divided by a characteristic depth of droplet penetration, hchar, or (4-3a) The rate of dissolution of dispersed oil per volume of seawater is proportional to the number of droplets per volume (~Q/hchard3) times the surface area of a drop (~d2). Hence the concentration of dissolved oil (4-3b) A simple model for the characteristic depth is hchar ~ Ez/ws, where Ez ~ hw2 (independent of depth), and ws ~ d2 (from Eq. 4-2). The wave flume experiments by Delvigne and Sweeney (1988) suggest that Q ~ hw1.14, while d is independent of hw. Thus, from Eq. (4-3a), cdrop ~ d2/hw0.86, and from Eq. (4-3b), cdiss ~ d/hw0.86. With this “model” both droplet and dissolved phase concentrations decrease with wave height and increase with droplet diameter. In reality, diffusivity is not likely to be constant with depth so an alternative model assumes a characteristic depth that is proportional to wave height, or hchar ~ hw. In this case, equations (4-3a) and (4-3b) give cdrop ~ hw1.14 and cdiss ~ hw1.14/d. Here both droplet and dissolved phase concentrations increase with wave height and either decrease with, or are independent of, droplet diameter, i.e., quite different from the conclusions of the first model. These arguments are qualitative, and more precise information should come from computer models that integrate multiple mechanisms in a quantitative manner as later discussed. But computer models are no better than our understanding of the individual mechanisms upon which they are based, and the uncertainty in even the direction of change noted above suggests we need better understanding of dispersant effectiveness (i.e., the dependence of Q and d on oil properties and environmental parameters), as well as better models of the vertical distribution of Ez, in order to accurately predict the concentrations of dispersed oil. Horizontal Subsurface Transport Subsurface advection of dispersed and dissolved phase oil by a uniform current affects the location of the oil, but does not, in itself, cause
OCR for page 142
Oil Spill Dispersants: Efficacy and Effects additional mixing. However, mixing is produced when the currents are non-uniform, and this mixing is responsible for the ultimate dilution of the oil. Without horizontal mixing, and under sufficiently calm weather conditions, vertically dispersed oil droplets could all ultimately resurface given enough time. Horizontal mixing consists of two fundamental processes. The first process is called scale-dependent diffusion and represents the fact that large eddies will advect a patch of marked fluid if the patch is smaller than the scale of the eddies, but mix and dilute the patch if it is larger than the eddies (Csanady, 1973). The second process is termed shear dispersion and results from the combination of velocity gradient(s) in combination with mixing (or other transport mechanism) in the direction of the gradient(s) (Fischer et al., 1979). The latter effect is enhanced with the use of chemical dispersants, because the smaller droplets that are produced are transported deeper, where they experience greater differences in horizontal velocity. Unfortunately, field measurements cannot always distinguish the two processes, and frequently their effects are combined. Horizontal mixing is determined best using site-specific measurements, but as these are often not available, literature values should be used. Okubo (1971) summarizes a number of coastal tracer studies and shows that (4-4) where σr is a characteristic radius (standard deviation) of an equivalent circular tracer patch (cm) and t is time (sec). Other investigators report similar trends. Okubo’s data apply to patch sizes ranging from ~30 m to ~100 km, and more recent data suggest the approximate relationship applies to even larger scales (Ledwell et al., 1998). Simple relationships such as this are useful because dilution resulting from horizontal mixing is proportional to patch variance, σr2, and hence Eq. (4-4) can be used to directly compute changes in concentration due to horizontal mixing. Also, predictive models make use of horizontal diffusion coefficients (Er, with dimensions of L2/T) defined by the time rate of change of patch variance. For example, using Eq. (4-4) (4-5) For σr = 100 m, Er = 0.3 m2/s, while for σr = 1,000 m, Er = 5 m2/s. Note that these values of horizontal diffusivity are orders of magnitude larger than the corresponding vertical values (Ez) suggesting that horizontal mixing is much stronger than vertical mixing. However, horizontal mixing is also much less effective, because horizontal plume dimensions are much larger
OCR for page 143
Oil Spill Dispersants: Efficacy and Effects (and hence horizontal concentration gradients are much smaller) than in the vertical. It should also be recognized that different investigators define horizontal diffusion coefficients differently. For example, as implied above, some data used to determine mixing coefficients include the effects of vertical shear, while others do not. Also some analyses separate Er into separate components in the longitudinal and lateral direction (i.e., an Ex and Ey), and some analyses define an apparent diffusivity based on a cumulative, rather than instantaneous, change in σr2 (i.e., Era = σr2/4Δt). In order for a predictive model not to over or under account for mixing, care should be taken to define Er in the same way in the model that it was defined in the analysis of field measurements used to determine its value. Horizontal mixing can be considered important to the dilution process when it has caused the patch concentration to be diluted by a significant amount. Again using Eq. (4-4), the time required for patch size to increase from σr to σr (a two-fold increase in dilution) is (4-6) where Δtdouble is in sec, and σr is in cm. For example, Δt = 12 hours for σr = 1,000 m, and only about 1.7 hours for σr = 100 m. The fact that this time increases with σr suggests that horizontal mixing is more important for small spills, and that dispersants can be used more effectively when applied before substantial spreading has occurred (i.e., small σr). Of course, other factors affecting dispersant effectiveness are also time dependent. Tank studies, or small-scale field experiments, cannot be used to directly simulate horizontal mixing because the spills in such tests are too small, and there are additional artifacts due to the presence of walls. While horizontal mixing data such as those compiled by Okubo (1971) usually include the effects of shear dispersion, it is interesting to consider this component separately and evaluate how it varies with sea state and dispersant effectiveness. One type of shear dispersion that was discussed previously involves larger droplets that become vertically entrained into the water column and later rise to the surface. Because the slick generally travels faster than the underlying water, the droplets will re-enter the slick at the “back-of-the-pack,” leading to a long tail. This effect can be especially important nearshore, where vertical circulation is more pronounced. Indeed, this effect has been proposed as the reason oil from the Braer spill off the Shetland Islands was observed to travel in the opposite direction of the surface current (Proctor et al., 1994; Ritchie and O’Sullivan, 1994; Spaulding et al., 1994). Smaller droplets that are (nearly) permanently dispersed, and hence behave like water, are also affected by conventional shear dispersion. Con-
OCR for page 144
Oil Spill Dispersants: Efficacy and Effects sider a parcel of marked seawater occupying a depth hchar. If there is a vertical gradient in the near-surface velocity of magnitude λz (dimensions of velocity per depth, or T−1), the patch will experience a top-to-bottom velocity difference of Δu = λzhchar. Following Taylor’s analysis of longitudinal dispersion (see Fischer et al., 1979), a shear dispersion coefficient Esd (part of Er) ~ (Δu)2hchar2/Ez ~ λz 2hchar4/Ez. Based on the previous discussion of hchar, Esd is expected to increase strongly with increasing wave height and decreasing droplet diameter, suggesting an increase in shear-induced mixing, and hence dilution of dispersed oil, as sea state and dispersant effectiveness increase. The above discussion clearly implies that the vertical dimension needs to be included in modeling the transport of dispersed oil—not just to represent the concentration field, but also to properly represent the velocity field (i.e., a model needs to realistically represent the vertical gradients in velocity). Normally this requires a 3-D model. In shallow water, dispersed oil may become distributed over the entire water depth. However, even in this case, vertical gradients in velocity are important for dispersing the oil and these gradients should be accounted for, either by explicitly simulating the vertical shear in a 3-D model, or by computing horizontal shear dispersion coefficients for use in a 2-D (depth-integrated) model. In deeper locations where the dispersed oil is not uniformly distributed over depth, the oil will tend to be concentrated in a relatively thin horizontal layer near the surface. As with models of thermal or salinity stratification, this horizontal layering can present numerical challenges associated with resolving strong near-surface gradients. Resolution can be enhanced by employing models with stretched coordinates, such as σ-coordinates (that use a constant number of vertical grid cells regardless of water depth) or γ-coordinates (that, in addition, provide unequal grid spacing, allowing greater resolution near the surface). However, care should be taken to minimize or counteract the spurious vertical mixing that may result with such models due to the fact that the “horizontal” grid lines are not parallel with the stratification (Huang and Spaulding, 1995). FATE AND WEATHERING In addition to spreading and drift as discussed earlier, there are numerous processes that affect the ultimate fate of spilled oil or petroleum products (Figure 4-1). These include evaporation, dissolution, dispersion of whole oil droplets into the water column (entrainment), interaction of dissolved and dispersed components with suspended particulate material (SPM), photooxidation, biodegradation, uptake by organisms, water-in-oil emulsification (mousse formation), and stranding on shorelines (NRC, 1985, 1989, 2003).
OCR for page 145
Oil Spill Dispersants: Efficacy and Effects Chapter 3 summarized the changes in rheological properties (viscosity, interfacial tension, density, etc.) that begin to occur immediately after oil is spilled. The changes in physical properties caused by water-in-oil emulsification are particularly important because they affect how spilled oil is physically dispersed (entrained) into the water column (with and without dispersants), the ability of oil spill skimmers to recover oil from the sea surface, the ability of pumps to transfer the collected oil, and the volume of collected material that requires storage and disposal. In the following sections, the chemical and physical changes to oil on the water surface (generally thought of as weathering) caused by evaporation, photooxidation, and water-in-oil emulsification are discussed, with particular emphasis given to the latter (including identification of the chemical constituents within oil that largely control emulsion behavior) because of its importance in controlling dispersant effectiveness. After that, the fate of physically and chemically entrained oil droplets in the water column is considered. In evaluating the fate of entrained oil droplets, the primary focus is on biodegradation of dispersant-treated oil and the interaction of both physically entrained and dispersant-treated oil droplets with suspended particulate material. Surface Oil Evaporation Weathering Evaporation of lower-molecular-weight volatile components from a surface slick is important for dispersant applications because it can indirectly affect the formation of stable water-in-oil emulsions through the precipitation of asphaltenes and resins that help to stabilize the emulsion (Fingas and Fieldhouse, 2003). As the solvent components are evaporated from the slick, these higher-molecular-weight components can precipitate to coat entrained water droplets in the emulsion and inhibit water-water droplet coalescence and phase separation (Sjoblom et al., 2003). In addition, the evaporative loss of mono-aromatic components (benzene, toluene, xylenes, etc.) and two- and three-ring polynuclear aromatic hydrocarbons (PAH) and their alkyl-substituted homologues can significantly reduce the toxicity of the oil and the concomitant water-soluble fractions generated after physical or chemically enhanced entrainment of oil droplets into the water column. Evaporation is the single most important and rapid of all weathering processes (McAuliffe, 1989), and it can account for the loss of 20–50 percent of many crude oils, 75 percent or more of refined petroleum products, and 10 percent or less of residual fuel oils (Butler, 1975; Butler et al., 1976; NRC, 1985; 2003). Most of the early studies on evaporation focused on the loss of individual hydrocarbon components as a function of their vapor pressures and other factors such as temperature, wind speed, and
OCR for page 182
Oil Spill Dispersants: Efficacy and Effects NOAA oil transport code 3-D GNOME (Simecek-Beatty et al., 2002). Hypothetical spills were simulated off the coast of Florida using a range of oils, wind speeds (that in turn affect surface transport, wave height, and horizontal and vertical mixing), percentage effectiveness of dispersant application, and oil droplet-size distribution. These models were selected because they are three-dimensional, they treat oil as a composite of pseudocomponents, and they have limited computing requirements to operate the models. They are also used in real time during oil spills by NOAA to provide scientific support to the FOSC. This review was not intended as an endorsement of these particular models and indeed similar sensitivity analysis should be conducted using other models. Results of some sensitivity runs are presented below while the entire set is contained in Appendix E. It is very difficult to predict where, when, and how much spilled oil moves. For example, the effect of wind on the movement of Alaskan North Slope crude oil is illustrated in Figure 4-10. This figure shows predicted changes in oil distributions 24 hours after oil is spilled on the water surface in the south Florida nearshore area under 2, 10 and 25m/s wind. As the wind becomes stronger, more oil is naturally entrained into the water column—from 0 volume percent at 2 m/s wind to 3 volume percent at 10 FIGURE 4-10 Predicted oil distributions 24 hours after the release of Alaskan North Slope crude oil (no dispersant applied) under 2-, 10-, and 25-m/s wind in nearshore off Florida Keys. There is no oil dispersed by a chemical dispersant for these three cases.
OCR for page 183
Oil Spill Dispersants: Efficacy and Effects m/s, to 31 volume percent at 25 m/s. The greater entrainment means that less oil floats on the water surface, and hence less is available for evaporation, a result that might seem counterintuitive. Oil concentrations in the water column vary depending on the amount of oil naturally dispersed (entrained), but they also reflect the diffusivity (which increases at higher wind speed) in the water column. This example indicates some of the complexity involved with the way that currents and wind, and in turn waves and diffusion, affect the horizontal and vertical movement of oil. When a dispersant is applied, more oil is entrained into the water column, and the droplets changes size distribution from the original size, further complicating oil transport and fate processes (see Figure 3-1). For example, assuming 50 percent effectiveness for a dispersant applied between 6 and 12 hours after the oil spill, 40 percent of the oil discussed previously was predicted to end up in the water column (37 percent by chemical entrainment and 3 percent by natural entrainment) under 10-m/s wind. Figure 4-11 shows the location of the predicted plume 24 hours after the spill. The oil spill location is marked by “+.” Black spots represent oil floating on the water surface, and the shaded areas show different ranges of oil concentrations in the top 1 m of the water column. The oil plume in the top 1 m of the water column is following a different trajectory at a different speed than the oil on the surface that is moved by the wind and the current. This figure also indicates that the area of the top 1 m of water column containing oil is about 64 km2, 2.5 times more than the contaminated top 1 m water area without dispersant application. Clearly such quantitative estimates would not be possible without a model. Further complexity comes from the fact, mentioned previously, that oil consists of a wide ranges of hydrocarbons. Although oil toxicity comes from the cumulative impacts of multiple hydrocarbon components, low-and intermediate-molecular-weight components such as BTEX and PAH tend to cause more acute risks to aquatic biota, as is discussed in Chapter 5. These components usually evaporate faster and to a greater extent than large-molecular-weight components such as wax, resins, and asphaltenes. The latter are contributing components in the formation of mousse, which makes it more difficult for a dispersant to work effectively (see Chapter 3). Table 4-2 presents Alaska North Slope crude oil’s chemical components (Environment Canada, 2005), as indicated in their distillation cuts (built into the ADIOS2 code), together with those of intermediate fuel oil (IFO) 300 and marine diesel oil used in the sensitivity analysis. As shown in this table, the Alaska North Slope crude oil has more low molecular-weight components than the two refined oils. Oil composition changes and emulsification occurring during the transport of spilled oil significantly alter the physical properties of oil, especially viscosity and dispersant effectiveness, as previously discussed. Thus, it is important to
OCR for page 184
Oil Spill Dispersants: Efficacy and Effects FIGURE 4-11 Predicted oil movement at 24 hours after the release of Alaskan North Slope crude at point + under 10-m/s wind with a dispersant application (additional details contained in the text of Chapter 4). simulate the behavior and transport of the components of the various hydrocarbons rather than treating oil as one substance. Figure 4-12 presents the predicted composition (a relative volume fraction of each distillation cut) of these three oils floating on the water surface 0 and 6 hours after the spill. Because of evaporation, the composi-
OCR for page 185
Oil Spill Dispersants: Efficacy and Effects TABLE 4-2 Distillation Cuts of the Three Oils Used in the Modeling Sensitivity Analysis Oil Cut Number Alaska North Slope Crude Oil Intermediate Fuel Oil 300 Diesel Fuel Oil Weight Fraction, wt percent Temperature, °C Weight Fraction, wt percent Temperature, °C Weight Fraction, wt percent Temperature, °C 1 1.0 42 1.1 180 1.1 120 2 4.0 98 1.1 200 1.1 140 3 5.0 127 6.4 250 1.1 160 4 5.0 147 9.4 300 3.2 180 5 5.0 172 7.2 350 5.2 200 6 10.0 216 8.1 400 20.4 250 7 10.0 238 6.0 450 31.9 300 8 5.0 247 3.0 500 25.5 350 9 5.0 258 4.9 550 9.7 400 10 5.0 265 9.8 600 1.0 450 11 5.0 272 14.7 650 — — 12 10.0 282 10.7 700 — — 13 30.0 >282 17.4 >700 — — SOURCE: Data from Environment Canada, 2005. tion of each oil changed significantly over time. Most of the cuts that distill at about 200° C [roughly 392° F] or lower, including alkanes with <10 carbons plus the monocyclic aromatics, benzene and toluene, ethylbenzene, o-, m-, and p-xylene, and most of the C2- and C3-substituted benzenes), evaporated within six hours. Thus, if a dispersant is applied six hours after the oil spill, it would not be expected to introduce these compounds into the water column. The IFO 300 does not naturally disperse into water due to its high viscosity (~15,000 cP), according to the modeling. On the other hand, diesel, with very low viscosity (~4 cP), disperses naturally (73 percent) without adding a chemical dispersant, and after 16 simulation hours, no diesel would be floating on the water surface. The combination of natural and chemical dispersal would disperse 78 percent of the diesel into the water, so there is no merit to applying a dispersant in this particular case. Because these refined oils have a low percentage of low-temperature distillation cuts, the IFO 300 and diesel evaporated only 10 and 18 volume percent, respectively, over 24 and 14 hours (much less than Alaska North Slope crude oil). This example clearly shows contrasting behavior of these three oils having different hydrocarbon composition.
OCR for page 186
Oil Spill Dispersants: Efficacy and Effects FIGURE 4-12 Predicted compositions of floating oils initially and 6 hours after the releases of Alaskan North Slope crude oil, Intermediate Fuel Oil (IFO) 300, and Marine Diesel Oil (additional details contained in the text of Chapter 4). These examples of the sensitivity analysis results illustrate very complex and sometimes competing interactions among oil types, environmental conditions, and dispersant use. Quantitative estimates of oil concentration distributions clearly require the use of computer models, especially those with oil pseudo-component modeling capabilities. A final motivation for the sensitivity study was to assess whether models could be used in real time to help decide whether or not to use chemical dispersants during an actual spill. These questions are particularly important in nearshore areas where the impacts of using—and not using—dispersants are likely to be most significant. Unfortunately nearshore areas are also the most complicated hydrodynamically. Although 3-D GNOME can accept a three-dimensional flow field, it presently uses two-dimensional flows that are calculated based on a simplified force balance involving pressure, Coriolis, bottom friction, and variation in water density adjusted by tide and wind. This simplified approach is justified because of the need to make simulations very quickly for real-time model predictions, and because field observations can be used to update model output. Because chemical dispersants help transport oil into the water column, realistic simulation of subsurface transport becomes more important when evaluating the use of chemical dispersants, and the same formulation may not be sufficient. It is recommended that a range of 3-D
OCR for page 187
Oil Spill Dispersants: Efficacy and Effects hydrodynamic formulations be evaluated with the goal of identifying approaches that are sufficiently accurate, yet still efficient, for real-time use. APPLYING KNOWLEDGE ABOUT THE TRANSPORT AND FATE OF DISPERSED OIL TO SUPPORT DECISIONMAKING As discussed in Chapters 2 and 3, many ultimate conclusions about the wise and effective use of dispersants in nearshore settings will need to be based on an accurate and adequate understanding of many processes controlling the transport and fate of dispersed oil. These processes may play a significant role from the instant the oil enters the environment, and they constrain a number of operational decisions and play a significant role in evaluating potential impacts of whole and dispersed oil on sensitive species or habitats. Fate and Weathering of Oil Oil on the Surface Better information is still needed to determine the window of opportunity and percent effectiveness of dispersant application for different oil types and environmental conditions. Coordinated research should be undertaken at bench and wave-tank scales to define those parameters that control oil dispersability as the oil is allowed to weather under carefully controlled but realistic environmental conditions. Overprediction of evaporation rates can be a problem with oil-weathering models that assume a well-mixed oil phase (which is probably only valid for very thin and relatively unweathered oil slicks) and also assume that resistance to mass transfer is entirely in the air phase. As a result, it may be inappropriate to always model oil as a well-mixed phase. Algorithms for both well-mixed and diffusion-controlled fluids may need to be sequentially utilized as a function of oil weathering-dependent viscosity changes to better approximate spilled oil evaporative behavior. Additional work is recommended to reconcile the differences between the empirical evaporation approach utilized by Fingas (1996, 1997, 1999a) and more traditional pseudo-component approaches as considered by Jones (1996, 1997), who has proposed a simplified pseudo-component (SPC) model relating molar volume, vapor pressure, and molecular weight to the boiling point of the components. Sediment Particle Interactions The ultimate fate of dispersed oil is poorly understood. Of particular
OCR for page 188
Oil Spill Dispersants: Efficacy and Effects concern is the fate of dispersed oil in areas with high suspended solids and areas of low flushing rates. Although this has been an area of recent research, there is still insufficient information on which to determine how chemically dispersed oil interacts with a wide variety of suspended sediment types, both short- and long-term, compared to physically dispersed oil. In this regard, there appears to be more information on short-term comparisons versus the longer-term fate of oil/SPM agglomerates generated with and without dispersant addition. In particular, the longer-term biodegradation of oil/SPM agglomerates in the water column has not been adequately studied. Likewise, there are uncertainties in how dispersed oil might be consumed by plankton and deposited on the seafloor with fecal matter or passed through the food chain. Relevant state and federal agencies, industry, and appropriate international partners should develop and implement a focused series of studies to quantify the weathering rates and final fate of chemically dispersed oil droplets in high SPM-concentration regimes compared with non-dispersed oil. Biodegradation Past research on the effects of dispersants on the biodegradation of petroleum hydrocarbons cannot be used to predict the fate of chemically dispersed crude oil at sea. The results of many of these studies may be confounded by metabolism of the dispersant or short-term effects of dispersants on bacterial attachment to oil droplets. When dispersed oil plumes become diluted by the transport processes that act in the surface layer of the ocean, however, the surfactants present in the dispersant will partition out of the oil into the surrounding seawater. If this partitioning is fast relative to the kinetics of bacterial attachment to oil droplets, the dispersant may not interfere with microbial uptake of the petroleum hydrocarbons (i.e., the dispersed oil droplets will behave like physically dispersed oil except the oil-water interfacial area will be larger due to entrainment of a larger number of small droplets in the water column). Therefore, future research on the kinetics of dispersed oil biodegradation should be conducted at low oil-water ratios to simulate conditions that represent those that follow significant dilution of the dispersed oil plume. In addition, the experimental designs of laboratory studies that have been used are probably inappropriate for estimating the in-situ biodegradation rate of oil that is floating on the sea surface, because the mixing energies that are typically applied are usually sufficient to result in substantial physical dispersion (i.e., oil droplets continuously break away from the floating slick and are entrained into the aqueous phase due to vigorous mixing) and there is little opportunity for formation of water-in-
OCR for page 189
Oil Spill Dispersants: Efficacy and Effects oil emulsions, which can dramatically reduce in-situ biodegradation rates. Therefore, the biodegradation rates for chemically dispersed and undispersed oil that have been compared in most laboratory studies are probably skewed in opposite directions relative to their in-situ rates: the biodegradation rates that have been measured in the laboratory for chemically dispersed oil are probably lower than what would prevail in a dispersed oil plume and those measured for undispersed oil are probably higher than could be realized in a floating oil slick that is not subject to a high degree of natural dispersion. Due to the difficulty of designing laboratory-scale experimental systems that adequately simulate the in-situ processes that are expected to affect the biodegradation rate of chemically dispersed oil, future biodegradation studies should be designed to support dispersed oil fate and transport modeling. Ideally, droplet-scale models of biodegradation kinetics should be developed and the appropriate kinetic parameters should be estimated. In general, existing oil biodegradation kinetics data cannot be used to support modeling of biodegradation in dispersed-oil fate and transport models, because one or more important variables (e.g., oil-water interfacial area, microbial population size, hydrocarbon concentrations as a function of time) were not monitored. Another major limitation for predicting the fate (and effects) of chemically dispersed oil based on available laboratory studies is that few studies have quantitatively investigated the biodegradation rates and products of compounds that are of most long-term concern. These include the high-molecular-weight PAH (e.g., 4- and 5-ring compounds), which are degraded slowly if at all by microorganisms, have the potential to bioaccumulate, and can exert chronic toxic, mutagenic, or developmental effects. Most studies have focused on bulk measurements of oil degradation (e.g., carbon dioxide production or reductions in TPH) or degradation of major components, such as n-alkanes. Although these are important metrics, because they measure the extent of reduction in the total oil mass, they may not be the most important drivers of long-term effects, because normal and branched alkanes are well known to be easily biodegradable by bacteria that are ubiquitous. So, while the rate of degradation of these compounds is of interest from a model mass-balance perspective, their ultimate fate is not in doubt. High-molecular-weight PAH, on the other hand, are likely to persist in the residual oil droplets, which may be ingested by animals in the water column or benthos where they can exert chronic effects. Therefore, the biodegradation kinetics and ultimate biotransformation products of high-molecular-weight PAH should be investigated using indigenous microbial communities from seawater. The ecological impact of these persistent compounds will be determined
OCR for page 190
Oil Spill Dispersants: Efficacy and Effects by their transport characteristics, which can best be predicted by accurate fate and transport models that include all relevant processes (including biodegradation) and robust estimates of the model parameters. Present and Possible Role of Models As discussed previously, various processes constrain a number of operational decisions and play a significant role in evaluating potential impacts of whole and dispersed oil on sensitive species or habitats. Models are, therefore, powerful and necessary tools to support decisionmakers during all phases of oil spill planning, response, and assessment. Currently trajectory analysis is a key component of contingency planning, real-time prediction of slick trajectory, size, and thickness, and in natural resource damage assessment. These models are not currently used in real time to support decisionmaking for dispersant use, but in principle they could be. The required sophistication of the models for these purposes varies, but their performance could be improved for all purposes. Specifically, they are incomplete in terms of their representation of the natural physical process involved, verification of the codes, and validation of the output from these models in an experimental setting or during an actual spill. Thus, their ability to predict the concentrations of dispersed oil and dissolved aromatic hydrocarbons in the water column with sufficient accuracy to aid in spill decisionmaking has yet to be fully determined. The sensitivity analysis identified that dispersant effectiveness and oil droplet size change are the most important parameters for dispersant application modeling. Unfortunately, oil spill models currently available do not simulate physical mechanisms and chemical reactions in order to predict these parameters. Emulsification is also an important process that greatly influences dispersant effectiveness. Predicting emulsification requires accurate oil properties, as well as conducting a detailed mechanistic investigation on emulsification processes and their influence on dispersant effectiveness. It is also important to evaluate turbulence in the open sea and reflect it more accurately in the transport and fate modeling. Models show significant progress for supporting real-time spill-response decisions regarding dispersants use, especially in complex nearshore regions; however, any improved models should be evaluated for their ability to satisfy this need. Oil trajectory and fate models used by relevant state and federal agencies to predict the behavior of dispersed oil should be improved, verified, and then validated in an appropriately designed experimental setting or during an actual spill. Specific steps that should be taken to improve the value of models for dispersant use decisionmaking include:
OCR for page 191
Oil Spill Dispersants: Efficacy and Effects Improve the ability to model physical components of dispersed oil behavior (e.g., shear in vertical dimension, distribution of horizontal velocities as a function of depth, variations in the vertical diffusivity as a function of depth, sea-surface turbulence, etc.) Improve the ability for models to predict concentrations of dissolved and dispersed oil, expressed as specific components or pseudocomponents, that can be used to support toxicity analysis Validate how advective transport of entrained oil droplets is modeled through specifically designed flume/tank studies and open-ocean (spill of opportunity) tests. Develop an ability to predict the formation of water-in-oil emulsions under a variety of conditions Conduct a sensitivity analysis based on three-dimensional, oil-component, transport and fate models, and develop necessary databases (evaporation, dissolution, toxicity, etc.) for the oil-component based assessment approach Once the models are improved, they will be valuable tools for transport and fate modeling and associated biological assessments with and without dispersants. They should be used as part of the overall effort to define operational guidelines for dispersant use, including what oils are dispersible and for how long, the predicted effectiveness of dispersant application (which will be a key input into predicting the dispersed oil concentrations in the water column), likely extent and duration of different oil concentrations of concern, and guidelines for buffers around sensitive resources. Because this study did not explicitly evaluate the pseudo-components and their dissolved chemical components of the oil in the water column with and without dispersant application, additional sensitivity analyses should be conducted with three-dimensional oil-component transport and fate models. It is also important to develop the necessary database (evaporation, dissolution, toxicity, etc.) for the pseudo-component-based assessment approach. This evaluation focused more on nearshore water, and it is recommended to also conduct sensitivity modeling for offshore, semi-confined waters and rivers. A consensus regarding “how good is good enough” needs to be developed among decisionmakers and model developers, and used to guide the future development of models and to optimize their use in real time. In discussions with NOAA modelers, it was noted that predicting the three-dimensional flow distribution as a part of the oil transport and fate modeling within several hours after an oil spill is difficult. A real-time model application uses actual environmental conditions and oil properties, but, because of time limitations, uses simple approaches for approxi-
OCR for page 192
Oil Spill Dispersants: Efficacy and Effects mating hydrodynamic data. To reflect three-dimensional flow and mixing, NOAA is implementing simple schemes to handle vertically varying diffusion and horizontal velocity fields. There have been some attempts to incorporate surface flow measurements into real-time oil transport models (Hodgins et al., 1993; Ojo and Bonner, 2002). However, these require pre-installation of data acquisition (e.g., high frequency radar) and transmission systems, and are currently applicable only to horizontal surface current and diffusion with relatively coarse grid resolution—not for the three-dimensional distributions needed for the three-dimensional modeling (Ojo and Bonner, 2002). The growing availability of ocean observing systems in coastal waters will likely improve the availability of real-time data useful for improved modeling of physical processes. Unlike real-time model applications, a pre-planning assessment uses hypothetical environmental conditions and oil properties, but can use detailed models, including complex three-dimensional flow fields. Thus, real-time and pre-planning modeling efforts should complement each other to provide better information to a decisionmaker. One of the greatest weaknesses in correlating laboratory-scale and mesoscale experiments with conditions in the open ocean derives from a lack of understanding the turbulence regime in all three systems. Likewise, one of the biggest uncertainties in computer modeling of oil spill behavior (with and without dispersant addition) comes from obtaining appropriate horizontal and vertical diffusivities. It is difficult to integrate all interacting transport and fate processes and oil properties to predict how much oil will be found in specific areas during an actual oil spill without the use of models. Relevant state and federal agencies, industry, and appropriate international partners should develop a coordinated program to obtain needed information about turbulence regimes at a variety of interrelated scales. This effort should include a field program to measure the upper sea-surface turbulence, under a variety of conditions with particular emphasis on quantifying horizontal and vertical diffusivities and the rate of energy dissipation, which can be compared to similar turbulent regimes in mesocosm systems.
Representative terms from entire chapter: