How Do Oil, Gas, and Water Interact Near a Subsea Blowout?

Home , Subsea (technology)

University of South Florida Scholar Commons

C-IMAGE Publications C-IMAGE Collection

8-2016

Scott A. Socolofsky Texas A&M University

E. Eric Adams Massachusetts Institute of Technology

Claire B. Paris-Limouzy University of Miami

Di Yang University of Houston

Follow this and additional works at: https://scholarcommons.usf.edu/cimage_pubs

Part of the Marine Biology Commons

Scholar Commons Citation Socolofsky, Scott A.; Adams, E. Eric; Paris-Limouzy, Claire B.; and Yang, Di, "How Do Oil, Gas, and Water Interact Near a Subsea Blowout?" (2016). C-IMAGE Publications. 67. https://scholarcommons.usf.edu/cimage_pubs/67

This Article is brought to you for free and open access by the C-IMAGE Collection at Scholar Commons. It has been accepted for inclusion in C-IMAGE Publications by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. OceTHE OFFICIALa MAGAZINEn ogOF THE OCEANOGRAPHYra SOCIETYphy

CITATION Socolofsky, S.A., E.E. Adams, C.B. Paris, and D. Yang. 2016. How do oil, gas, and water interact near a subsea blowout? Oceanography 29(3):64–75, http://dx.doi.org/10.5670/ oceanog.2016.63.

DOI http://dx.doi.org/10.5670/oceanog.2016.63

USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: [email protected] or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA.

DOWNLOADED FROM HTTP://TOS.ORG/OCEANOGRAPHY GoMRI: DEEPWATER HORIZON OIL SPILL AND ECOSYSTEM SCIENCE

How Do Oil, Gas, and Water Interact Near a Subsea Blowout?

By Scott A. Socolofsky, E. Eric Adams, Claire B. Paris, and Di Yang

64 Oceanography | Vol.29, No.3 At low flow rates into seawater, oil jets are laminar and break up into droplets having nearly uniform diameters… However, for higher flow rates, the“ jet is turbulent, and the oil becomes atomized into a spectrum of smaller droplets whose diameters decrease with increasing flow rate. .

ABSTRACT. Oil and gas from a subsea blowout shatter into droplets and bubbles that was near complete dissolution of methane,” rise through the water column, entraining ambient seawater and forming a plume. Local significant dissolution of small hydrocar- density stratification and currents eventually arrest this rising plume, and the entrained bon molecules (Ryerson et al., 2011), and water, enriched with dissolved hydrocarbons and some of the smaller oil droplets, near complete oxidation of methane in forms one or more subsurface intrusion layers. Beyond the plume and intrusion the water column (Du and Kessler, 2012). layer(s), droplets and bubbles advect and diffuse by local currents and dissolve and These measurements highlight the impor- biodegrade as they rise to the surface, where they are transported by wind and waves. tance of subsurface processes to hydro- These processes occur over a wide range of length scales that preclude simulation by carbon fate, yet there remain significant any single model, but separate models of varying complexity are available to handle uncertainties in the processes responsible the different processes. Here, we summarize existing models and point out areas of for these effects. ongoing and future research. To bridge this gap, studies of transport processes in the nearfield (where INTRODUCTION small-scale processes that transport oil vertical plume rise is significant) and far- The Deepwater Horizon (DWH) acci- and gas near a subsea blowout. field (where bubbles and droplets advect dent demonstrated in dramatic fashion Most observations made during DWH independently) are underway by sev- the wide range of processes that affect oil were beyond the 5 km response zone— eral groups, many funded by the Gulf of droplets and gas bubbles released from hence, beyond the region of major droplet/ Mexico Research Initiative (GoMRI). To a subsea blowout. These processes begin bubble dynamics—and primary observa- communicate research quickly among with breakup of the blowout jet into small tions were of the spill’s dissolved signature this network of colleagues, the Nearfield droplets and bubbles, followed by their within the water column (e.g., Valentine, Modeling Listserv was created (near- vertical transport as a buoyant plume, et al., 2010; Kessler, et al., 2011; Du and [email protected], mod- horizontal transport within intrusion lay- Kessler, 2012). As in all spills, droplet size erated by author Socolofsky), and the ers caused by local density stratification, distribution is critical to predict the oil’s user group has hosted three work- advection and diffusion by local currents, fate and transport. For DWH, significant shops. This group identified five areas for and additional mixing near the surface quantities of chemical dispersants were study: (1) bubble and droplet generation, due to wind and waves. Simultaneously, applied subsurface, directly at the spill (2) plume modeling, (3) intrusion forma- dissolution and biodegradation alter the source, to promote formation of smaller tion and coupling to circulation models, oil and gas mixtures. How these pro- oil droplets (Brandvik et al., 2013). (4) particle tracking models, and (5) bub- cesses occur, their rates, and what man- However, no measurements of bubble ble and droplet fate modeling (an inte- agement strategies may affect them are or droplet size distributions were made grative topic present in each of the pre- important questions as we look to under- in situ at the source. The few measure- vious four areas). Here, we discuss the stand the impact of DWH on the Gulf ments made inside the response zone con- first four topics and touch on dissolution of Mexico and to seek mitigation strate- firmed the strong plume behavior of the (topic five), showing how oil and gas mix gies for potential future oil spills. Here, oil and gas jet (Camilli et al., 2011), deter- and are dispersed from a subsea blowout. we discuss our present understanding mined the emission composition (Reddy, For farfield transformation, see Tarr et al. of, and the ongoing research addressing, et al., 2011), and demonstrated that there (2016, in this issue).

Oceanography | September 2016 65 MODELING DEEPWATER atomized into a spectrum of smaller observed median droplet diameters d50, BLOWOUTS droplets whose diameters decrease with measured in laboratory experiments with Droplet Generation increasing flow rate (Figure 1). The tran- oil jetted into seawater (e.g., Brandvik The sizes of bubbles and droplets gener- sition from laminar to turbulent jet et al., 2013), to the Weber number using ated in a subsea blowout affect hydrocar- breakup depends on a Weber number the orifice diameter and velocity as bon transport and fate in several ways. (Tang and Masutani, 2003), length and velocity scales in Equation 1. First, oil and gas are buoyant, causing the ρU2 D The simplest equation in this empirical We =–0 , (1) hydrocarbons to rise toward the surface as σ approach is a plume (see section on Nearfield Plume where ρ = oil density, U = exit velocity, d 0 –50 = AWe –3/5 (2) Dynamics, below). Second, depending D = orifice diameter, and σ = interfacial D , on their sizes, the droplets and bubbles tension. At the time of DWH, no consen- where A is a fitting coefficient. Following may separate from the entrained seawater, sus models were available to predict actual Wang and Calabrese (1986), Johansen resulting in different pathways to the droplet sizes in the atomization regime. et al. (2013) modified Equation 2 to surface (see section on Intrusion Layer Under these energetic conditions, tur- account for viscosity (which resists drop- Formation and Farfield Tracking Models, bulent pressure fluctuations cause oil to let breakup when interfacial tension is below). And third, droplets and bubbles break up into increasingly smaller drop- reduced due to the use of dispersants). provide the surfaces across which hydro- lets until they reach a critical size at which They also accounted for the presence of carbon constituents are dissolved into sea- breakup is resisted by interfacial tension gas mixed with the oil. For a given oil flow water and ultimately degraded. Chemical (Brandvik et al., 2013). Coalescence due rate and orifice, the gas increases the exit dispersants, designed to reduce interfacial to droplet collision may also play a role. velocity of oil; it also increases the down- tension and, thus, droplet size, override For decades, chemical engineers have stream buoyancy of the jet relative to a this process. The effectiveness of disper- studied such processes under equilib- pure oil jet. Johansen et al. (2013) thus sants depends on how much smaller they rium conditions, and developed correla- predict median droplet size as a function can make the droplets, and how much dif- tions of characteristic droplet size with of a modified Weber number, ferently smaller droplets are transported a Weber number, choosing values for U0 –3/5 d50 We compared with larger, undispersed drop- and D appropriate for a stirred reactor – = A –– , (3) D 1 + BVi (d /D) lets. While both droplets and bubbles are (Hinze, 1955). However, except for a short [ 50 ] important, we focus here on oil droplets. distance of several orifice diameters from where We is now based on Uc = a corrected At low flow rates into seawater, oil jets the source, oil emanating from a blowout exit velocity to account for gas and buoy- are laminar and break up into droplets is not in equilibrium, but instead exhibits ancy, Vi = µUc /σ, µ = dynamic viscosity of having nearly uniform diameters, com- rapidly decreasing turbulence and droplet oil, and A and B are empirical coefficients. parable to that of the orifice or the max- concentration along the jet trajectory. While the effect of oil viscosity has been imum stable droplet size, whichever is Two basic approaches have been taken well verified by laboratory experiments, smaller. However, for higher flow rates, to predict droplet size under the dynamic the effect of gas has not, partly due to the the jet is turbulent, and the oil becomes conditions in a jet. The first calibrates difficulty in distinguishing gas bubbles

FIGURE 1. Oil droplets discharged into water through a 2 mm diameter nozzle (Tang and Masutani, 2003). From left to right, the jet exit velocities are 0.31, 0.97, 1.41, 2.42, and 5.99 m s–1. As a point of reference, the nozzle outer diameter is 2.5 cm.

66 Oceanography | Vol.29, No.3 from oil droplets when both are present. interfacial tension and oil viscosity in most models predicted droplet sizes Equation 3 predicts mean droplet sizes resisting breakup and are coupled with of 1–10 mm without dispersants, and that agree well with a wide range of lab- nearfield buoyant jet models that com- 0.1–1 mm if dispersants were uniformly oratory experiments and one small- pute spatially varying turbulence prop- mixed with the oil at a DOR of 1:50, scale field study—DeepSpill described erties and droplet concentrations. Unlike though there was significant variability in Johansen et al. (2003)—and provides the empirical equations described above, among predictions. Indeed, remaining a promising method to extrapolate to population models simulate the entire questions being addressed with ongoing the scale of a major blowout. Aman et al. droplet size distribution and suggest that experiments concern the effects of using (2015) developed a variant of this equa- this distribution evolves with distance, a live oil (containing dissolved gas) and tion by implicitly assuming equilibrium facet beginning to be verified experimen- dependence on temperature and pres- conditions within the jet, and determin- tally. As with the correlation equations, sure. Moreover, in other spill scenarios ing model coefficients using droplet sizes both population models agree well with with different initial conditions, mechan- observed with oil stirred in a reactor. The available laboratory and field data, but ical breakup may be adequate without the form of their empirical equation is sim- only after calibration. use of dispersants. Nonetheless, for the ilar, but their fit coefficients give drop- Laboratory experiments (Gopalan conditions tested, model predictions gen- lets sizes that are two orders of magni- and Katz, 2010; Nagamine, 2014) show erally agree that the use of chemical dis- tude smaller than those of Johansen et al. that when chemical dispersants are persants reduces both droplet sizes and (2013). The Johansen equation has been introduced, additional (latent) breakup corresponding droplet rise velocities, shown to predict reliably against labo- occurs far from the jet release due to pro- resulting in more than a tenfold increase ratory data and one small field exper- cesses such as tip-streaming and tearing. in the downstream length of the surfac- iment involving jet breakup, but data Nagamine (2014) observed time-varying oil footprint—a significant measure at large scale are lacking to fully vali- ing behavior of droplets held in place of the effectiveness of subsea injection of date either model’s predictions. using a counter-flowing arrangement. chemical dispersants. Empirical equations such as Centimeter-size droplets impregnated Equations 2 and 3 only predict charac- with chemical dispersants at disper- Nearfield Plume Dynamics teristic droplet sizes (e.g., d50) and must sant-to-oil ratios (DOR) > 1:250 broke Gas bubbles and oil droplets released assume a distribution for droplet sizes into very fine droplets (1–50 μm) within from a blowout exert an upward force on around the characteristic size. Typically, minutes, while similar size droplets with the local water column, rising collectively either a Rosin-Rammler or a lognor- lower DOR remained stable for over a day. as a buoyant, multiphase plume. The mal distribution is used (Johansen et al., At a recent model intercomparison plume entrains ambient seawater, lifting 2013), with parameters describing the workshop, modelers compared their pre- it to higher elevations in the water col- distribution widths taken from labo- dictions of droplet size and transport umn. As the plume rises, the gas bubbles ratory experiments. These models also (Socolofsky et al., 2015). For a spill size and smaller molecules of the oil droplets require specification of the interfacial approximately one-third that of DWH, dissolve into the entrained seawater. This tension σ. If the oil is untreated, σ should be known, but for oil treated with dispersants, the effectiveness with which the dispersant and oil mix depends on the amount, method, and location of injec- FIGURE 2. Schematic of a tion, causing uncertainty in the actual dynamic (population) model. interfacial tension and, hence, the pre- The population of droplets of a given size (represented dicted droplet size (Brandvik et al., 2013; by the red and green drop- Socolofsky et al., 2015). let in the center) may increase through the breakup of larger The other basic approach to deter- droplets or the coalescence of mine droplet size uses a dynamic (or smaller droplets, while the pop- population) model (Figure 2) to simu- ulation may decrease through the breakup or coalescence of late droplet breakup and coalescence as droplets of the given size. oil encounters time-varying conditions. Notable population models developed for multiphase plumes include VDROP-J (Zhao et al., 2014) and Nissanka and Yapa (2016). Both account for effects of

Oceanography | September 2016 67 reduces plume buoyancy as well as bub- Stratification and crossflow in buoy- integral models based on the entrainment ble and droplet diameters, thereby affect- ant multiphase plumes have been stud- hypothesis have been successfully applied ing their rise velocities, and increases ied in the laboratory and in the DeepSpill to predict multiphase plume dynam- the concentration of dissolved hydro- field experiment. Key parameters con- ics (Milgram, 1983). Two major blow- carbons in the plume. Eventually, local trolling their dynamics include: wr = bub- out simulation models are DeepBlow density stratification and currents cause ble/droplet rise velocity, B = kinematic (Johansen, 2003) and Clarkson Deep Oil the entrained seawater to fall out of the buoyancy flux of dispersed phase parti- and Gas (CDOG; Zheng and Yapa, 2002). plume, or detrain, forming lateral intru- cles, N = buoyancy frequency of ambient These models have been carefully vali- sion layers of enhanced hydrocarbon stratification, and ua = ambient current dated and can predict the dissolution of content, which were observed more than velocity (Socolofsky and Adams, 2002, gas bubbles (Zheng and Yapa, 2002), usu- 100 km downstream of the DWH blow- 2005). (BN)1/4 is a characteristic veloc- ally treating oil droplets as inert. They run out (Kessler et al., 2011; Du and Kessler, ity (Socolofsky and Adams, 2005), and quickly, making them ideal for response 3 2012). The plume width and intrusion B/wr gives a characteristic length scale of and for exploring sensitivity to complex layer thickness span scales up to a few a multiphase plume (Bombardelli et al., bubble and droplet behavior and chem- hundred meters, much smaller than the 2007). Combinations of these parameters istry. Disadvantages are that they can- resolution of ocean circulation models; have been used to predict characteristics not resolve unsteady flow features or the hence, these features are simulated using of plumes (Figure 3). complex processes of detrainment, intru- submodels specifically designed to cap- Although self-similarity is not sion formation, and weak plume dynam- ture fine-scale dynamics. strictly valid for multiphase plumes, ics above the detrainment point. Recently, large eddy simulation (LES) has been applied to complex oil and gas plumes in stably stratified conditions (Fabregat et al., 2015, and recent work of Alexandre Fabregat, City University Subsequent Plume of New York, and colleagues; Fraga et al., 2016; Yang et al., 2016). LES models do not rely on self-similarity and are Q2 able to directly resolve large- and intermediate-scale turbulent motions, rely- Peeling Region ing on closure models for the effects of subgrid-scale (SGS) features. E These LES models treat water as the p Outer λ1bi Plume continuous phase and bubbles and drop- Peel height (hP) lets as dispersed phases. Using an Eulerian bi Ea framework, the incompressible Navier- Stokes equations are solved for the water velocity field, and a convection-diffusion E o equation solves for the water density field, Ei which is coupled to the buoyancy term Qi in the Navier-Stokes equations using the Boussinesq approximation (Fabregat Inner Plume et al., 2015; Yang et al., 2016). Both flow Trap height (h ) T Ei and density equations are filtered at the LES grid scale, and several different SGS models have been used to close the equations. For example, Fraga et al. (2016) used the standard Smagorinsky FIGURE 3. Schematic of a multiphase plume discharging to a stratified quiescent water column (after Socolofsky et al., 2008, with permission from ASCE). Bubbles and droplets create SGS model with a constant coefficient;

a buoyant plume that rises, entraining ambient seawater along the way. At some elevation hP, Fabregat et al. (2015, and recent work the mixture of relatively light bubbles and droplets plus heavy entrained seawater runs out of of Alexandre Fabregat, City University momentum, and the seawater detrains (peels), ultimately becoming trapped at an elevation of New York, and colleagues) employed hT where it is neutrally buoyant relative to the ambient seawater. Symbols relate to the double (inner and outer) plume model used to describe this phenomenon. a spectral vanishing viscosity technique

68 Oceanography | Vol.29, No.3 suitable for their spectral-element code; the contributions from all particles in thermal/gas hybrid plumes. By consid- and Yang et al. (2016) used a Lagrangian- the vicinity of an Eulerian grid point. ering varying bubble rise velocities, Yang averaged scale-dependent dynamic SGS By contrast, Fabregat et al. (2015, and et al. (2016) systematically studied instan- model developed for complex turbulent recent work of Alexandre Fabregat, City taneous and mean plume characteris- flows with spatial inhomogeneity. University of New York, and colleagues) tics, focusing on the turbulent entrain- Regarding dispersed phases, LES and Yang et al. (2016) employ Eulerian– ment, peeling, and intrusion processes models can be categorized into two Eulerian models, defining a dispersed (Figure 4a). Yang et al. (2016) also assessed approaches. In Fraga et al. (2016), phase Eulerian concentration, which the flux parameterizations typically used Eulerian-Lagrangian models track the obeys a convection-diffusion equation. in integral plume models. Based on their motions of individual particles based Using the Eulerian–Eulerian approach, analysis, they proposed a new continuous on Newton’s second law according to Fabregat et al. (2015, and recent work peeling model for double-plume integral forces acting on individual particles. of Alexandre Fabregat, City University models with more self-consistent perfor- The reaction force from the particles to of New York, and colleagues) studied mance than previous models. the fluid is calculated by accounting for the characteristics of thermal, gas, and A major challenge for both integral

θ (K)

–1 a 283 284 285 w (m s ) b –0.03 0.00 0.03 0.6

0 0.4 z (m) –100

z (m) 0.2 400 –200

200 –300 y (m) 0 0.0 400 –0.2 0.0 0.2 200 –200 x (m) 0 x (m) –400 –200 0.0000 0.0001 Dye concentration (kg m–3) –400

400 c d

10–1 200

–2

10 ) 0 –3 z (m)

10–3 (kg m –200 10–4 Surface oil mass concentration –400

–400 –200 0 200 400 –400 –200 0 200 400 x (m) x (m) FIGURE 4. Sample results obtained by the Eulerian–Eulerian large eddy simulation (LES) model. (a) Laboratory-scale LES of nearfield plume (corresponding to the case WR6 in Yang et al., 2016), taken after 80 s of simulated bubbling on the x-z plane, across the center of the source, with contours of instantaneous dye concentration. (b–d) Field-scale LES of farfield oil plume dispersion in an ocean mixed layer

for a flow condition with Langmuir number Lat = 0.43 (corresponding to the flow case L2 in Yang et al., 2015). (b) A plume with oil droplets 250 microns in diameter, with contours of vertical velocity w and temperature θ shown on the two vertical planes at x = 500 m and y = 500 m, respectively. (c) Surface oil mass concentration (kg m–3) for a plume with 500-micron oil droplets. (d) Surface oil mass concentration (kg m–3) for a plume with 125-micron oil droplets. Details of the simulation setups for the nearfield and farfield LESs can be found in Yang et al. (2016) and Yang et al. (2015), respectively.

Oceanography | September 2016 69 and computational fluid dynamics (CFD) Intrusion Layer Formation Using Equation 4 for an oil with den- models is predicting dissolution of oil The previous section describes nearfield sity of 0.85 g cm–3, droplets with a diame- droplets and gas bubbles. Dissolution is plume dynamics, showing how buoy- ter of 2–20 mm, typical of those expected complicated by the complex chemistry of ant oil and gas, released at the bottom of for untreated oil at DWH (Testa et al., in hydrocarbon mixtures (Chen and Yapa, a stratified ocean, can become trapped press), would be transported 50–100 m 2001), including the possible formation in layers, centered on the level of neu- radially within the intrusion layer. of clathrate hydrate skins on the bubbles tral buoyancy of the entrained seawater. Droplets that are 0.2–2 mm in diameter, or droplets. Hydrate formation is pres- It is of interest to know whether oil drop- typical of those for dispersant treated oil sure- and temperature-dependent, and lets also become trapped. Experimental without any latent breakup (Zhao et al., has recently been studied in high-pres- studies suggest the classification shown 2015), would travel 100–500 m later- sure laboratory facilities (Warzinski et al., in Figure 5 (Chan et al., 2015), indi- ally. Finally, droplets with a diameter of 2014) and at natural seeps (B. Wang cating that, as the characteristic veloc- 0.02–0.2 mm, typical of those expected 1/4 et al., 2016). In the latter study, in situ ity UN = wr /(BN) decreases, droplets at DWH for dispersant-treated oil expe- high-speed imagery of seep bubbles near become more effective at pumping ambi- riencing significant latent breakup 1,000 m depth confirmed the formation of ent water upward to one or more intru- (Nagamine, 2014) could theoretically be hydrate skins on gas bubbles from natural sion layers, and there is greater tendency transported 3–20 km before rising out of seeps in the field, and field measurements for droplets to detrain and enter the intru- the intrusion layer. These transport dis- indicate that mass transfer rates vary sion themselves. Chan et al. (2014) found tances are in general agreement with pre- between rates for clean and dirty bubbles a threshold value of UN = 0.2 to 0.4, below dictions for similar droplet sizes in far- (Rehder et al., 2009; B. Wang et al., 2016). which droplets intrude. They also derived field transport models (Paris et al., 2012; Mass transfer rates are maintained at the theoretically and confirmed experimen- North et al., 2015). dirty bubble rate (i.e., they are not further tally a relationship for the distance σr that reduced by the hydrate skin), likely due to a droplet travels within the first intrusion Farfield Tracking Models cracks on the hydrate skin, as observed by layer before escaping, given by We have seen how hydrocarbons from Warzinski et al. (2014), and the rise height a blowout break up into small oil drop- of the seep flares depends on the rise of 0.9 – 0.38(U )0.24 B3/8 lets and gas bubbles, and how these buoy- –––N –– σr = 5/8 1/2 . (4) the largest gas bubbles released from the π N Wr ant fluids interact in the nearfield with seep. More measurements are needed to local ocean currents and stratification determine the hydrocarbon distribution Additional experiments have been con- to form a plume and intrusion layers. in the water column, and to understand ducted to establish similar threshold val- For the farfield, Paris et al. (2013) con- how the greater turbulent mixing of live ues of UN and lateral transport distances sider mixing of rising oil droplets and oil and gas in a real blowout may differ for oil discharging into a mildType current 1a* dissolved hydrocarbons,Type and 1b* Lagrangian from that in a weak natural seep flare. (Wang and Adams, 2016). stochastic models (LSM) can be used to UN < 0.3 0.3 < UN < 1.4

Type 1a* Type 1b* Type 2 Type 3

UN < 0.3 0.3 < UN < 1.4 1.4 < UN < 2.4 UN > 2.4 FIGURE 5. Plume classification scheme (after Chan et al., 2015, reproduced with the permission of Springer). The four panels depict droplet plumes with four ranges of dimensionless droplet slip velocity UN given by the ratio of the droplet rise velocity to a characteristic water velocity in the plume. As UN decreases, droplets rise more slowly and become more effective at pumping seawater upward. They are also more passive in the plume, and are broadcast more widely in the radial direction.

70 Oceanography | Vol.29, No.3

Type 2 Type 3

1.4 < UN < 2.4 UN > 2.4 simulate subsequent fate and transport the temperature and salinity of the ambi- of droplets as they rise through the water of multi-fraction Lagrangian elements ent water, diffused by ambient turbu- column. The smallest droplets surface (Paris et al., 2012; North et al., 2015). lence, and transformed by a host of phys- farthest from the source, and mitigation The locations, flow rates, and composi- ical, chemical, and biological processes, measures to reduce droplet size (e.g., sub- tions of oil droplets leaving the intrusion including dissolution, biodegradation surface dispersant injection) or more layers become initial conditions for far- (Lindo-Atichati et al., 2014), high pres- energetic breakup regimes can cause the field transport models. These inputs can sure (Aman et al., 2015), and sediment surface expression of the oil to move be obtained from integral or LES models particle interactions (Paris et al., 2012; downstream relative to the unmitigated of the nearfield, as discussed previously, or North et al., 2015). While oil transforma- case (Socolofsky et al., 2015). from correlation equations, as reported in tions are critical for determining oil fate, The dispersion term used in farfield Socolofsky et al. (2011). In the LSM, par- we focus here on transport and mixing. tracking models is normally taken from ticle sizes are typically selected from the For a discussion of transformation, see the horizontal and vertical turbulent dif- droplet size distribution (DSD) at the end Tarr et al. (2016, in this issue). fusivities computed in the ocean circu- of the nearfield, either by binning the data Ocean circulation models, such as lation models. These diffusivities can be (North et al., 2015) or by a random num- SABCOM (used with the transport spatially variable and are treated with ber generator matched to the DSD proba- model LTRANS; North et al., 2015) and random walk algorithms, as in the DWH bility function (Paris et al., 2012); particle GoM-HYCOM (used with the oil appli- hindcasts of North et al. (2015) and Paris properties, including the density of var- cation of the Connectivity Modeling et al. (2012). Alternatively, model diffu- ious hydrocarbon fractions, are matched System, CMS; Paris et al., 2013) provide sivities can be estimated from field mea- to the results of the nearfield model. the velocity fields, and transport mod- surements. For example Ledwell et al. The boundary conditions specify- els calculate the local currents by fine- (2016) measured concentrations of an ing the three-dimensional flow field scale interpolation from the CFD grid- introduced tracer (SF5CF3) to deter- come from ocean circulation models. ded velocity. These CFD models generate mine scale-dependent vertical diffu- Currently, the coupling between the near- three-dimensional flow fields over com- sivities, and Z. Wang et al. (2016) used field plume and ocean circulation mod- plex bathymetry, often using nested a microstructure profiler to determine els is only in one direction (i.e., the near- domains that provide added resolution in small scale turbulent properties, both field model depends on the farfield flow critical areas, and may respond to tides, near the DWH site. field but not vice versa). This one-way coupling is also true between the farfield tracking and ocean circulation models. Because the nearfield plume induces significant vertical velocity and the intru- The sizes of bubbles and droplets generated sion layer can generate a large flow rate in a subsea blowout affect hydrocarbon transport (about 1,000 m3 s–1 for the primary intrusion during DWH), two-way coupling and fate in several ways. . may be important between the buoyancy- “ dominated near and intermediate field (region of the intrusion formation) and the ocean circulation. The CARTHE and C-IMAGE II consortia are study- wind, air/sea fluxes, density variations, One important property affecting” par- ing this problem, but to date, no two- and Earth’s rotation (Coriolis force), ticle transport is the turbulent velocity 1/2 1/2 way coupled hindcast simulations for the among others. Many ocean circulation we = ε /N , where ε = the turbulent dis- near- and farfield of the DWH accident models also rely on data assimilation to sipation rate. By comparing we with the have been reported. keep model predictions on track. rise velocity of oil droplets of varying den- For an LSM, oil is typically repre- For tracking of oil droplets, the rise sity and diameter, Z. Wang et al. (2016) sented by Lagrangian elements represent- velocity of the droplets, which is depen- conclude that droplets larger than 0.4 ing dissolved hydrocarbons or oil drop- dent on droplet diameter and density mm (rise velocity 6 mm s–1) are unlikely lets with appropriate droplet size and (Zheng and Yapa, 2000), is added to the to be significantly affected by turbu- density (Paris et al., 2012; North et al., advection predicted from the ocean cur- lence, while those smaller than 0.04 mm 2015). These Lagrangian elements are rents. Because smaller droplets rise more (rise velocity 0.1 mm s–1) are expected advected by mean ocean currents and slowly than larger droplets, horizontal to become so entangled with turbulence droplet rise velocity that varies based on currents stretch the spatial distribution that they might be permanently trapped

Oceanography | September 2016 71 below the surface. interaction may be warranted. dispersion term in the tracking model. In addition to contributing to drop- Figure 6 shows a result from the CMS Hence, results for farfield transport mod- let diffusion, turbulence can also alter model for DWH using GoM-HYCOM els should always be interpreted with droplet rise velocity. Bellani and Variano for the ocean circulation. The figure visu- respect to the resolution and assumptions (2012) used a novel underwater imaging alizes the farfield oil distribution by iso- in both the underlying ocean circulation technique to compare effects of turbu- surfaces of oil concentration for July model and the coupled Lagrangian track- lence on particles and vice versa, focus- 14, 2010, the day the Macondo well was ing model. ing on spherical particles (characteristic capped. We chose high concentration of small oil droplets) and prolate ellipsoi- values to illustrate the complex effects Processes in the Surface dal particles (characteristic of larger oil of planetary rotation and bathymetry on Mixed Layer droplets). Their results show that smaller hydrocarbons in the water column near After being subjected to turbulent droplets are more affected by turbulence, the DWH blowout; the full extent of the entrainment and peeling/intrusion pro- leading to slower rise velocities than in subsurface distribution of droplets would cesses in the nearfield and lateral trans- the quiescent case. Current transport be seen if lower concentration thresholds port processes in the farfield, some of the models use correlations for terminal rise were included. The figure shows the anti- oil droplets finally rise to the very upper velocity in quiescent ambient conditions. cyclonic behavior of the oil due to plane- portion of the ocean—the ocean mixed For large droplets, these correlations are tary rotation in the GoM-HYCOM sim- layer (OML)—where physical proper- likely acceptable, but for smaller drop- ulation. Large eddies are captured in the ties of seawater are well mixed. Important lets, these correlation equations will pre- velocity field predicted by HYCOM, illus- flow structures, for example, wind-gen- dict oil surfacing closer to the source than trating the importance of the underlying erated turbulence, surface waves, Stokes may actually occur. For response model- circulation model predictions for the far- drift, Langmuir circulations, and Ekman ing, this behavior is conservative, but for field tracking models for long-term sim- spirals, often coexist in the OML, caus- effects modeling and damage assessment, ulations. Smaller eddies not captured in ing lateral and vertical dispersion and more sophisticated droplet-turbulence the circulation model are folded into the affecting the oil’s surface footprint. Oil

FIGURE 6. Planetary effect on the Deepwater Horizon blowout. Oil structure in the farfield is visualized by the isosurfaces of oil concentration for 550, 110, and 25 ppb on day 85 (July 14, 2010), the day prior to capping of the wellhead. The visualization does not include the nearfield plume. The isosurfaces selected here are for high concentrations to enhance visualization of the anticyclonic rotation; the smaller oil concentrations showing the extent of the spill are not represented here.

72 Oceanography | Vol.29, No.3 reaching the surface from a deep spill simultaneously the effects of Langmuir (2016b) developed a new approach called will generally form a very thin slick, espe- circulation, turbulence, Stokes drift, the Extended Nonperiodic Domain cially if chemical dispersants have been and oil droplet buoyancy (Figure 4b–d). LES for Scalar transport (ENDLESS), used. Thus, its subsequent spreading is Their results reveal that instantaneous oil which offers several major advances: not amenable to classical analyses applied plume patterns as well as the averaged (1) It increases the effective LES domain to surface spills that involve balances of plume transport direction are affected by size by solving the flow field on a typi- gravity, inertia, viscosity, and surface ten- two competing mechanisms, downward cal periodic LES domain and tracking the sion (e.g., Hoult, 1972). Instead, other mixing induced by Langmuir turbulence, evolution of the nonperiodic oil plume factors dominate the spreading, includ- characterized by the velocity US, and the field over a laterally extended domain ing: (1) oil droplets take different path- droplet rise, given by wr, summarized by (replicating the LES velocity field peri- ways to the surface, which increases their the ratio Db = US /wr. Plumes with large odically). (2) It efficiently tracks large- surface footprint; (2) wind, waves, and droplets (Db < 10) tend to form fingered scale plume dispersion by adaptively add- non-uniform currents diffuse and break patterns, with the mean transport primar- ing and removing simulation blocks for up the slick; and (3) droplets periodi- ily downwind (Figure 4c); plumes with the scalar solver, depending on the real- cally disperse and resurface, thus tending small droplets (Db > 25) tend to be highly time spatial extension of the oil plume. to stretch the plume, a process likely to diffused, with significant crosswind mean (3) It incorporates the numerical capabil- increase in significance when considering transport (Figure 4d); and plumes with ity of superimposing larger-scale quasi- the fate of chemically dispersed oil. Many intermediate droplets (10 < Db < 25) usu- two-dimensional flow motions on oil of these flow phenomena have character- ally have blurred patterns that exhibit advection, allowing for coupling with istic length scales much smaller than the surface convergence and form fingered regional circulation models. As the first submesoscale, so they are not considered patterns, but with no clean gaps between step, this new ENDLESS model may serve in typical large-scale ocean circulation fingers (Figure 4b). to bridge the gap between traditional LES models used in predicting oil transport Based on an extensive set of LES runs, and large-scale ocean circulation models. (e.g., SABGOM and HYCOM). Yang et al. (2015) studied the behavior Recently, LES-based ocean turbulence of eddy viscosity and eddy diffusivity for CHALLENGES AND models have been used to study fine-scale oil dispersion, and proposed a modified OPPORTUNITIES oceanic flows and their effects on scalar K-profile parameterization (KPP) that Nearfield behavior spans wide spatial dispersion. Using LES, Yang et al. (2014, incorporates the effects of Langmuir cir- scales, from submillimeter droplet-scale 2015) and Chen et al. (2016b) studied culations and oil droplet buoyancy into processes to advection by ocean cur- oil dispersion in ocean Langmuir turbu- classical KPP. Subsequently, Chen et al. rents over hundreds of kilometers, and lence, a flow system that combines the (2016a) also applied the LES model to it depends on intricate multiphase fluid aforementioned fine-scale flow phenom- study the effect of arbitrarily oriented dynamics, chemistry, and biodegrada- ena in the OML. Their LES model solved ocean swell waves on the transport and tion of complex mixtures at extreme the filtered Craik–Leibovich equations, dilution of oil in the OML. They found temperatures and pressures. Much suc- which comprise a wave-averaged version that when varying the swell-wind relative cess has been achieved through labora- of the Navier–Stokes equations but differ angle, strong swell waves not only affect tory, field, and numerical experimenta- from the regular Navier-Stokes equations the mean transport of the oil plume via the tion, yet many uncertainties remain. A by inclusion of an extra vortex force term strong swell-induced Stokes drift but they major challenge affecting the initial con- that accounts for the cumulative effects also influence instantaneous oil dilution dition for transport simulations is that oil of surface waves on the shear turbulence by modulating the intensity of the turbu- droplet breakup occurs at larger We than responsible for generating Langmuir cir- lence and the Langmuir circulations. what can be achieved in laboratory-scale culations. The intensity of the vortex The aforementioned LES studies illus- models. Present predictions of droplet force can be measured by the turbulent trate the importance of considering fine- size for DWH, for instance, must extrap-

Langmuir number Lat = √u* /US , where scale ocean flow structures in order to olate several orders of magnitude beyond u* = the wind-induced friction velocity in accurately predict mean-plume trans- the nearest measurement. Because of the OML and US = the magnitude of the port and dilution, which are typically the importance of droplet size on down- Stokes drift. Scalar quantities obey reg- not included in regional ocean circula- stream processes, scaled-up experiments ular transport equations, which include tion models due to limited grid resolu- are critically needed. the wave-induced Stokes drift. tion. On the other hand, the typical LES At the small scale, uncertainties Using their LES model, Yang et al. domain size of order 1 km is insufficient include the sensitivity of model predic- (2014, 2015) studied the complex disper- for tracking long-range plume evolu- tions to complex chemical and biological sion of oil plumes in the OML, capturing tion. To overcome this limit, Chen et al. processes. Dissolution and mass transfer

Oceanography | September 2016 73 are fundamentally known, and recent Bombardelli, F.A., G.C. Buscaglia, C.R. Rehmann, Johansen, Ø., H. Rye, and C. Cooper. 2003. L.E. Rincon, and M.H. Garcia. 2007. Modeling DeepSpill: Field study of a simulated oil and gas data are helping to understand the role and scaling of aeration bubble plumes: blowout in deep water. Spill Science & Technology of hydrates. However, in computationally A two-phase flow analysis. Journal of Hydraulic Bulletin 8:433–443, http://dx.doi.org/10.1016/ Research 45(5):617–630, http://dx.doi.org/10.1080/ S1353-2561(02)00123-8. expensive models, like LES and the far- 00221686.2007.9521798. Kessler, J.D., D.L. Valentine, M.C. Redmond, field tracking models, expensive chem- Brandvik, P.J., Ø. Johansen, F. Leirvik, U. Farooq, M. Du, E.W. Chan, S.D. Mendes, E.W. Quiroz, and P.S. Daling. 2013. Droplet breakup in sub- C.J. Villanueva, S.S. Shusta, L.M. Werra, and others. istry calculations may not be affordable. surface oil releases: Part 1. Experimental study of 2011. A persistent oxygen anomaly reveals the In this case, it remains to understand the droplet breakup and effectiveness of dispersant fate of spilled methane in the deep Gulf of Mexico. injection. Marine Pollution Bulletin 73(1):319–326, Science 331:312–315, http://dx.doi.org/10.1126/ sensitivity of model predictions to these http://dx.doi.org/10.1016/j.marpolbul.2013.05.020. science.1199697. processes and to develop simplified rep- Camilli, R., D. Di Iorio, A. Bowen, C.M. Reddy, Ledwell, J.R., R. He, S.F. DiMarco, L. Spencer, and A.H. Techet, D.R. Yoerger, L.L. Whitcomb, P. Chapman. 2016. Dispersion of a tracer in the resentations of oil fate. J.S. Seewald, S.P. Sylva, and J. Fenwick. 2011. deep Gulf of Mexico. Journal of Geophysical At the large scale, there remains a Acoustic measurement of the Deepwater Research 121:1,110–1,132, http://dx.doi.org/ Horizon Macondo well flow rate. Proceedings 10.1002/2015JC011405. need to couple the nearfield dynam- of the National Academy of Sciences of the Lindo-Atichati, D., C.B. Paris, M. Le Hénaff, ics with ocean circulation models in United States of America 109:20,235–20,239, M. Schedler, A.G. Valladares Juárez, and R. Müller. http://dx.doi.org/10.1073/pnas.1100385108. 2014. Simulating the effects of droplet size, order to resolve submesoscale eddies Chan, G., A. Chow, and E.E. Adams. 2015. Effects of high-pressure biodegradation, and variable flow and account for the effect of Earth’s rota- droplet size on intrusion of sub-surface oil spills. rate on the subsea evolution of deep plumes Environmental Fluid Mechanics 15:959–973, from the Macondo blowout. Deep Sea Research tion. In the case of DWH, the oil mix- http://dx.doi.org/10.1007/s10652-014-9389-5. Part II 129:301–310, http://dx.doi.org/10.1016/ ture released for 86 days was subject to Chen, B., C. Yang, C. Meneveau, and M. Chamecki. j.dsr2.2014.01.011 2016a. Effects of swell on transport and dispersion Milgram, J.H. 1983. Mean flow in round bubble Coriolis forcing, which affected the far- of oil plumes within the ocean mixed layer. Journal plumes. Journal of Fluid Mechanics 133:345–376, field dynamics (Figure 6). Several studies of Geophysical Research-Oceans 121:3,564–3,578, http://dx.doi.org/10.1017/S0022112083001950. http://dx.doi.org/10.1002/2015JC011380. Nagamine, S.I. 2014. The effects of chemical dis- in the C-IMAGE and CARTHE consortia Chen, B., D. Yang, C. Meneveau, and M. Chamecki. persants on buoyant oil droplets. MSc Thesis, show the importance of these processes 2016b. ENDLESS: An extended nonperiod Department of Mechanical Engineering, University domain large-eddy simulation approach for of Hawaii at Manoa. on mixing and flow stability, but integrat- scalar plumes. Ocean Modelling 101:121–132, Nissanka, I.D., and P.D. Yapa. 2016. Calculation ing these effects with field-scale models http://dx.doi.org/10.1016/j.ocemod.2016.04.003. of oil droplet size distribution in an under- Chen, F.H., and P.D. Yapa. 2001. Estimating hydrate water oil well blowout. Journal of Hydraulic remains a challenge. The inexpensive inte- formation and decomposition of gases released Research 54(3):307–320, http://dx.doi.org/10.1080/ gral models are steady state; hence, they in a deepwater ocean plume. Journal of Marine 00221686.2016.1144656. Systems 30:21–32, http://dx.doi.org/10.1016/ North, E.W., E.E. Adams, A.E. Thessen, Z. Schlag, can quickly adapt to changing bound- S0924-7963(01)00032-X. R. He, S.A. Socolofsky, S.M. Masutani, and ary conditions, but they may not be effec- Du, M.R., and J.D. Kessler. 2012. Assessment S.D. Peckham. 2015. The influence of drop- of the spatial and temporal variability of let size and biodegradation on the transport of tive at predicting coupled dynamics. LES bulk hydrocarbon respiration following the subsurface oil droplets during the Deepwater models show promise in this area, and the Deepwater Horizon oil spill. Environmental Horizon spill: A model sensitivity study. Science & Technology 46(19):10,499–10,507, Environmental Research Letters 10(2), 024016, ENDLESS approach described above may http://dx.doi.org/10.1021/es301363k. http://dx.doi.org/10.1088/1748-9326/10/2/024016. yield a working model. But, presently, Fabregat, A., W.K. Dewar, T. Özgökmen, and Paris, C.B., J. Helgers, E. van Sebille, and A.C. Poje. 2015. Numerical simulations of turbu- A. Srinivasan. 2013. Connectivity Modeling continued improvement in model reso- lent thermal, bubble and hybrid plumes. Ocean System: A probabilistic modeling tool for the lution and in determining the sensitivity Modelling 90:16–28, http://dx.doi.org/10.1016/ multi-scale tracking of biotic and abiotic vari- j.ocemod.2015.03.007. ability in the ocean. Environmental Modelling of model predictions to these dynamics Fraga, B., T. Stoesser, C.C.K. Lai, and S.A. Socolofsky. & Software 42:47–54, http://dx.doi.org/10.1016/ is required. Meanwhile, advances in cou- 2016. A LES-based Eulerian–Lagrangian approach j.envsoft.2012.12.006. to predict the dynamics of bubble plumes. Ocean Paris, C.B., M.L. Hénaff, Z.M. Aman, A. Subramaniam, pled nearfield-farfield dynamic modeling Modelling 9,727–9,736, http://dx.doi.org/10.1016/ J. Helgers, D.-P. Wang, V.H. Kourafalou, and together with the use of real-time, sev- j.ocemod.2015.11.005. A. Srinivasan. 2012. Evolution of the Macondo Gopalan, B., and J. Katz. 2010. Turbulent shearing of well blowout: Simulating the effects of the cir- en-day forecasts of high-resolution ocean crude oil mixed with dispersants generates long culation and synthetic dispersants on the sub- circulation (e.g., FKeyS-HYCOM) allow microthreads and microdroplets. Physical Review sea oil transport. Environmental Science & Letters 104, 054501, http://dx.doi.org/10.1103/ Technology 46:13,293–13,302, http://dx.doi.org/ for near-real-time tracking and forecast- PhysRevLett.104.054501. 10.1021/es303197h. ing of oil dynamics and seem to be the Hinze, J.O. 1955. Fundamentals of the hydro- Reddy, C.M., J.S. Arey, J.S. Seewald, S.P. Sylva, dynamic mechanism of splitting in disper- K.L. Lemkau, R.K. Nelson, C.A. Carmichael, most promising approach for rapid eval- sion processes. AIChE Journal 1(3):289–295, C.P. McIntyre, J. Fenwick, G.T. Ventura, and others. uation of blowout predictions to support http://dx.doi.org/10.1002/aic.690010303. 2011. Composition and fate of gas and oil released Hoult, D.P. 1972. Oil spreading on the sea. to the water column during the Deepwater first-response decisions. Annual Review of Fluid Mechanics 4:341–367, Horizon oil spill. Proceedings of the National http://dx.doi.org/10.1146/annurev.fl.04.010172. Academy of Sciences of the United States of REFERENCES 002013. America 109:20,229–20,234, http://dx.doi.org/ Aman, Z.M., C.B. Paris, E.F. May, M.L. Johns, and Johansen, Ø. 2003. Development and verification 10.1073/pnas.1101242108. D. Lindo-Atichati. 2015. High-pressure visual exper- of deep-water blowout models. Marine Pollution Rehder, G., I. Leifer, P.G. Brewer, G. Friederich, and imental studies of oil-in-water dispersion droplet Bulletin 47:360–368, http://dx.doi.org/10.1016/ E.T. Peltzer. 2009. Controls on methane bubble size. Chemical Engineering Science 127:392–400, S0025-326X(03)00202-9. dissolution inside and outside the hydrate stabil- http://dx.doi.org/10.1016/j.ces.2015.01.058. Johansen, Ø., P.J. Brandvik, and U. Farooq. ity field from open ocean field experiments and Bellani, G., and E.A. Variano. 2012. Slip velocity 2013. Droplet breakup in subsea oil releases: numerical modeling. Marine Chemistry 114:19–30, of large neutrally buoyant particles in turbulent Part 2. Predictions of droplet size distributions http://dx.doi.org/10.1016/j.marchem.2009.03.004. flows. New Journal of Physics 14:25,009–25,009, with and without injection of chemical disper- Ryerson, T.B., K.C. Aikin, W.M. Angevine, http://dx.doi.org/10.1088/1367-2630/14/12/125009. sants. Marine Pollution Bulletin 73(1):327–335, E.L. Atlas, D.R. Blake, C.A. Brock, F.C. Fehsenfeld, http://dx.doi.org/10.1016/j.marpolbul.2013.04.012. R.-S. Gao, J.A. de Gouw, D. Fahey, and others.

74 Oceanography | Vol.29, No.3 2011. Atmospheric emissions from the Deepwater on a methane bubble in water: Observations Horizon spill constrain air-water partitioning, and new insights for hydrate film models. hydrocarbon fate, and leak rate. Geophysical Geophysical Research Letters 41:6,841–6,847, Research Letters 38, L07803, http://dx.doi.org/ http://dx.doi.org/10.1002/2014GL061665. 10.1029/2011GL046726. Yang, D., M. Chamecki, and C. Meneveau. 2014. Socolofsky, S.A., and E.E. Adams. 2002. Multi- Inhibition of oil plume dilution in Langmuir phase plumes in uniform and stratified crossflow. ocean circulation. Geophysical Research Journal of Hydraulic Research 40(6), 661–672, Letters 41(5):1,632–1,638, http://dx.doi.org/ http://dx.doi.org/10.1080/00221680209499913. 10.1002/2014GL059284. Socolofsky, S.A., and E.E. Adams. 2005. Role Yang, D., B. Chen, M. Chamecki, and C. Meneveau. of slip velocity in the behavior of strati- 2015. Oil plumes and dispersion in Langmuir, fied multiphase plumes. Journal of Hydraulic upper-ocean turbulence: Large-eddy simula- Engineering 131(4):273–282, http://dx.doi.org/ tions and K-profile parameterization. Journal 10.1061/(ASCE)0733-9429(2005)131:4(273). of Geophysical Research 120(7):4,729–4,759, Socolofsky, S.A., E.E. Adams, M.C. Boufadel, http://dx.doi.org/10.1002/2014JC010542. Z.M. Aman, Ø. Johansen, W.J. Konkel, D. Lindo, Yang, D., B. Chen, S.A. Socolofsky, M. Chamecki, M.N. Madsen, E.W. North, C.B. Paris, and others. and C. Meneveau. 2016. Large-eddy simula- 2015. Intercomparison of oil spill prediction mod- tion and parameterization of buoyant plume els for accidental blowout scenarios with and with- dynamics in stratified flow. Journal of Fluid out subsea chemical dispersant injection. Marine Mechanics 794:798–833, http://dx.doi.org/10.1017/ Pollution Bulletin 96(1–2):110–126, http://dx.doi.org/ jfm.2016.191. 10.1016/j.marpolbul.2015.05.039. Zhao, L., M.C. Boufadel, E.E. Adams, S.A. Socolofsky, Socolofsky, S.A., E.E. Adams, and C.R. Sherwood. T. King, K. Lee, and T. Nedwed. 2015. Simulation 2011. Formation dynamics of subsurface hydrocar- of scenarios of oil droplet formation from the bon intrusions following the Deepwater Horizon Deepwater Horizon blowout. Marine Pollution blowout. Geophysical Research Letters 38, Bulletin 101(1):304–319, http://dx.doi.org/10.1016/ L09602, http://dx.doi.org/10.1029/2011GL047174. j.marpolbul.2015.10.068. Socolofsky, S.A., T. Bhaumik, and D.-G. Seol. 2008. Zhao, L., M.C. Boufadel, S.A. Socolofsky, E. Adams, Double-plume integral models for near-field mix- T. King, and K. Lee. 2014. Evolution of droplets ing in multiphase plumes. Journal of Hydraulic in subsea oil and gas blowouts: Development Engineering 134:(6)772–783, http://dx.doi. and validation of the numerical model org/10.1061/(ASCE)0733-9429(2008)134:6(772). VDROP-J. Marine Pollution Bulletin 83(1):58–69, Tang, L., and S.M. Masutani. 2003. Laminar and tur- http://dx.doi.org/10.1016/j.marpolbul.2014.04.020. bulent flow liquid-liquid jet instability and breakup. Zheng, L., and P.D. Yapa. 2000. Buoyant Paper presented at the Thirteenth International velocity of spherical and nonspheri- Offshore and Polar Engineering Conference, cal bubbles/droplets. Journal of Hydraulic May 25–30, 2003, Honolulu, HI. Engineering 126(11):852–854, http://dx.doi.org/ Tarr, M.A., P. Zito, E.B. Overton, G.M. Olson, 10.1061/(ASCE)0733-9429(2000)126:11(852). P.L. Adhikari, and C.M. Reddy. 2016. Zheng, L., and P.D. Yapa. 2002. Modeling gas disso- Weathering of oil spilled in the marine envi- lution in deepwater oil/gas spills. Journal of Marine ronment. Oceanography 29(3):126–135, Systems 31(4):299–309, http://dx.doi.org/10.1016/ http://dx.doi.org/10.5670/oceanog.2016.77. S0924-7963(01)00067-7. Testa, J., E.E. Adams, E.W. North, and R. He. In press. Estimating the influence of deep water applica- ACKNOWLEDGMENTS tion of dispersants on the surface expression This work was made possible by a grant from the Gulf of oil using a particle tracking model. Journal of of Mexico Research Initiative to the Gulf Integrated Geophysical Research. Spill Research (GISR) Consortium to the Center Valentine, D.L., J.D. Kessler, M.C. Redmond, for Integrated Modeling and Analysis of the Gulf S.D. Mendes, M.B. Heintz, C. Farwell, L. Hu, Ecosystem (C-IMAGE I and II), and to the RFP-II project F.S. Kinnaman, S. Yvon-Lewis, M.R. Du, and others. Large Eddy Simulation of Turbulent Dispersion of Oil 2010. Propane respiration jump-starts microbial in the Ocean Surface Layers: Development, Testing, response to a deep oil spill. Science 330:208–211, and Applications of Subgrid-Scale Parameterizations. http://dx.doi.org/10.1126/science.1196830. Wang, B., S.A. Socolofsky, J.A. Breier, and J.S. Seewald. 2016. Observations of bubbles in nat- AUTHORS ural seep flares at MC 118 and GC 600 using in Scott A. Socolofsky ([email protected]) is situ quantitative imaging. Journal of Geophysical Professor, Zachry Department of Civil Engineering, Research 121(4):2,203–2,230, http://dx.doi.org/ Texas A&M University, College Station, TX, USA. 10.1002/2015JC011452. E. Eric Adams is Senior Lecturer and Senior Research Wang, C.Y., and R.V. Calabrese. 1986. Drop breakup Engineer, Department of Civil and Environmental in turbulent stirred-tank contactors: Part II. Relative Engineering, Massachusetts Institute of Technology, influence of viscosity and interfacial tension. AIChE Cambridge, MA, USA. Claire B. Paris is Associate Journal 32(4):667–676, http://dx.doi.org/10.1002/ Professor of Ocean Sciences, Rosenstiel School aic.690320417. of Marine & Atmospheric Science, Miami, FL, USA. Wang, D., and E.E. Adams. 2016. Intrusion dynam- Di Yang is Assistant Professor, Department of ics of particle plumes in stratified water with Mechanical Engineering, University of Houston, weak crossflow: Application to deep ocean blow- Houston, TX, USA. outs. Journal of Geophysical Research 121, http://dx.doi.org/10.1002/2015JC011324. ARTICLE CITATION Wang, Z., S.F. DiMarco, and S.A. Socolofsky. Socolofsky, S.A., E.E. Adams, C.B. Paris, and D. Yang. 2016. Turbulence measurements in the north- 2016. How do oil, gas, and water interact near a ern Gulf of Mexico: Application to the Deepwater subsea blowout? Oceanography 29(3):64–75, Horizon oil spill on droplet dynamics. Deep Sea http://dx.doi.org/10.5670/oceanog.2016.63. Research Part I 109:40–50, http://dx.doi.org/ 10.1016/j.dsr.2015.12.013. Warzinski, R.P., R. Lynn, I. Haljasmaa, I. Leifer, F. Shaffer, B.J. Anderson, and J.S. Levine. 2014. Dynamic morphology of gas hydrate

Oceanography | September 2016 75