Oil-Mineral Flocculation and Settling Dynamics

Abstract

In recent decades, oil spill contamination has tended to occur more commonly in deltaic and estuarial systems. The management of oil spillages has been a major challenge in the surrounding deltas due to the highly sensitivity nature of deltaic ecosystems. Many deltas have an abundance of clay minerals that can flocculate, and these play an important role in determining the transport of spilled oil contamination and its eventual fate, particularly given that suspended sediment and microbial activities are often prevalent and diverse in natural environments. The primary work presented here focuses on laboratory experimental studies that help develop improved parameterizations of flocculation processes for oil-sediment-biogeochemical modeling. Oil-mineral flocs (OMA) have been successfully created from a series of laboratory flocculation experiments. A floc video instrument LabSFLOC-2 has been adopted for the first time to study the settling dynamics of OMAs. Experimental results reveal OMAs can easily form in any oil, cohesive sediment, and seawater mixtures. However, Kaolin and Bentonite forms dramatically different OMA structures, which leads to their variable characteristics. In the Bentonite clay cases, the oil flocs tend to be much larger and with higher densities than those in Kaolin clay cases, resulting in significant variability of flocs settling velocities.

1. Introduction

1.1 Scope of the problem

The rise in anthropogenic activities and industrialized related development in coastal and marine environments, and their ensuing recovery, from contamination by oil spillages, have been a considerable challenge [1]; especially in coastal deltaic regions where they have high vulnerability to aquatic ecosystem and associated public health issues [2]. During the past decades, contamination from oil spill incidents has gravitated to occur more frequently in vulnerable deltaic waters [3, 4]. To illustrate the global scale of the problem, a few examples of delta regions from around the world are presented.

In Africa, oil spill sites are a common phenomenon in the Niger Delta region. Between 1976 and 1996, a total of 4647 incidents resulted in the spilling of approximately 2,369,470 barrels of oil into the environment [5]. Of this quantity, an estimated 1,820,410.5 barrels (77%) were not recovered [5]. They tend to originate from a variety of sources including: leaks during processing, corrosion of oil pipes, poor maintenance of infrastructure, and deliberate acts of vandalism or theft of crude oil from pipes.

The Nigerian coastline is about 853 kilometers in length from the border with the Republic of Benin in the west to the Republic of Cameroon in the east. Encompassing an area of approximately 70,000 km², the Niger Delta is one of the largest wetlands in the entire world. The region comprises: sandy and muddy morphological features, swamp forests (both seasonal and permanent), low-land rain forests, and mangroves—both saline and brackish [6]. The region is topographically characterized by numerous rivulets, creeks, canals, and rivers of various sizes. The coastal line borders the Atlantic Ocean and is subjected to direct tidal interaction. In contracts, the mainland experiences episodes of flooding from the riverine networks, dominated by the River Niger. The soils underlying the Niger delta are generally characterized as soft, highly compressible, organic, and inorganic silty clays overlying fine sands at great depths [7].

Historically, the largest oil spill in Nigeria occurred during January 1980, when an offshore well blew out and oil spread throughout Nigeria’s Atlantic coastline. This spill caused extensive damage to nearly 340 hectares of mangrove [5] and was equivalent to approximately 200,000 barrels of crude oil (this is equivalent to 8.4million US gallons). Although not fully verified, over the past half century, it is estimated that oil spill incidents account for nearly 546 million gallons of oil [8] into the Niger Delta environment. This is comparable to approximately 11 million gallons per year [9]. UNEP [10] reported that many communities in the Niger Delta, in particular Ogoniland region, are continuing to live with a persistent state of pollution from oil spills.

Oil spills have blighted a myriad of the Nigeria Delta environment including: the ambient air, ground and surface waters, and crops (via bioaccumulation). Ordinioha and Brisibe [8] disclose that associated contaminants can include (but not limited to): trace metals; hydrocarbons, such as carcinogens, namely polycyclic aromatic hydrocarbon and benxo (a) pyrene; and naturally occurring radioactive substances.

For example, the Bodo oil spill in the Niger Delta region in 2008, which was caused by operational problems, recorded approximately 4000 barrels of oil spill a day for a 10-week period. The scale of the spill is likened to the Exxon Valdez 1989 disaster in Alaska, where 10 million gallons of oil destroyed the remote coastline [5].

The Niger Delta has ecologically sensitive wetlands, which makes the impact of oil spills to be widely felt by the people in the region. The people depend on the environment for traditional livelihood (farming and fishing) and subsistence; thus the contamination of water, land and air quality continuously erode their capacity to support the lifestyle and economy survival [9, 11]. When spills occur, whether on lands or waters caused by operational error, sabotage and theft, equipment failure, and aging pipeline, it contaminates water use for domestic purpose, fishing and farming activities with severe impact on the local people who are left with no alternative.

In the United Kingdom, since 1924, the Firth of Forth has received effluent from Scotland’s major petrochemical industry and refinery in addition to hydrocarbon inputs from many other sources [12]. This has created a large residue of hydrocarbon contaminants within its sediments. The major inputs have produced localized lethal effects, in that productive estuarine intertidal habitat has been lost or its functioning altered. The changes are primarily due to organic enrichment with only very near-field toxic effects. Some of the effects result from historical contamination, but existing discharges continue to have a deleterious effect. New petrochemical discharges in both firth and estuary appear to have a minimal impact, although the historical contamination makes this partly inconclusive [12].

Of particular note, the largest oil spill event in human history (at the time of publication) occurred in 2010 and was the Deepwater Horizon (DWH) disaster. This occurred in the vicinity of numerous coastal delta zones, and a significant net influence on: human activities, ecosystem contaminations, and environmental changes. Of the habitats recurringly oiled throughout the DWH spill event, Martin [13] reported that salt marshes were among the most frequently affected (45%), with remedial activities occurring on less than 9% of the affected territories [14]. The resilience of these salt marsh habitats is vital to the persistence of the native fish species that populate these marsh regions, together with the vivacity that is redirected to pelagic food webs [15]. The vulnerability of estuarine ecosystems and their fauna to oil released from DWH has been illustrated by numerous studies (e.g. [16, 17, 18, 19]).

DWH released approximately 4.9 million barrels (or 779 million L) of crude oil into the Gulf of Mexico [20, 21]. Michel et al. [14] estimated that the DWH spill directly affected approximately 1773 km of the shoreline habitat (much of this deltaic). In addition to the bleak ecological impacts the spill site imposed on neighboring shoreline, there was an unanticipated sedimentation of oil-associated marine snow (often referred to as Marine Oil Snows - MOS) throughout the water column, extending down to the seafloor; this gave rise to a supplementary prolonged impact on benthic zones [22, 23, 24].

Once oil is spilled into an aquatic environment, it tends to float as it is less dense (i.e., more buoyant) than the surrounding water, and this connection with the overlying atmosphere can facilitate the natural weathering processes on the oil contaminants. Having said that, a number of studies [23, 25] observed that a significant amount of the floating oil droplets readily aggregated with suspended matter already present in the water column, including (but not limited to): fecal pellets, detritus, sediments, and phytoplankton. In addition, both Passow and Alldredge [26] and Malpezzi et al. [27] found that this aggregation process was markedly enhanced by the presence of biologically secreted substances specifically Transparent Exopolymer Particles (TEP) or the more generic Extracellular Polymeric Substance (EPS; e.g. [28, 29, 30]). These MOS would eventually settle and deposit on the seafloor as they were transported through the water column [22, 31].

Following an oil spill, chemical dispersants are frequently chosen to remedy visible surface contaminants [32, 33, 34]. However, in the wake of dispersing an oil slick into smaller constituent oil droplets, there is a likelihood that these diminutive oil droplets agglomerate with other cohesive suspended materials present in the water column to further augment MOS flocculation and settling [35, 36]. Many deltaic regions, including estuarine and coastal water systems with energetic flows [37, 38, 39], are often dominated by muddy sediments comprising clay-based minerals and EPS (such as those exuded by phytoplankton). When oil is introduced into the sedimentary matrix, Zhao et al. [40] report that they can freely flocculate into oil-particle aggregates (OPAs). Ye et al. [41] concluded that this interchange between cohesive minerals, oil contaminants, and biological EPS collectively perform a pronounced role in the fate of oil spills in a natural ecosystem. Thus, an improved understanding of the formation, physical characteristics, and settling dynamics of OPAs is extremely advantageous when assessing how they will affect a subaquatic ecosystem [42, 43, 44].

Apart from the large amount of spilt oil floating at the water surface, there is still a considerable portion of oil being transported within the water column and settling to the seafloor after/during flocculating with natural sediment (especially clay mineral in deltaic regions) and biological materials (including EPS). And the interactions of oil and aquatic particles (sediment and biological materials) can be one of the keys for the spilt oil bio-degradation, oil-weathering or geological deposition, and resuspension processes. However, the flocculation processes of the different types of natural clay minerals when combined with oil are still poorly understood; especially their resultant settling dynamics.

From the 2010 Deepwater Horizon event, for example, round 3–5% spilt oil settling into the seafloor and still 11–25% spilt oil still uncounted for transporting or settling in the ocean water column (e.g., [20, 21]). The aggregation and settling processes can potentially lead to oil eventually residing in a quasi-preserved status within the seafloor sedimentary deposits, and these oil-contaminated sediments having an extended residence time. As such, this problem triggered this study, which not only addresses the improvement of our poor understanding of the multiple oil-mineral structures but also the oil-mineral flocculation processes and its resultant settling dynamics under certain turbulent flow conditions.

Through a series of well-controlled laboratory experiments, our goal is to improve our understanding of the settling velocity and floc properties associated with the flocculation of oil-contaminated clay minerals. This laboratory research is part of the Consortium for Simulation of Oil-Microbial Interaction in the Ocean (CSOMIO) with an overarching goal to advance the numerical model predictions for the future oil spill mitigation.

1.2 Overviews of oil-mineral aggregates (OMAs) studies

Oil-mineral aggregates (OMAs) are oil droplets stabilized by fine mineral particles in water. Since the mineral particles and oil interactions in aquatic system were first mentioned by Poirier and Thiel in 1941, for the few past decades numerous studies have been examining how oil hydrocarbons absorb onto mineral particles at a molecular level, (e.g., [45, 46, 47]). More recently, most oil-mineral aggregates (OMAs) studies have been focused on the: formation mechanisms [48, 49], characterizations [50], and “influence factors” such as salinity [51], temperature [50], oil types [50], and temporal turbulence [52]. Most of these past OMAs studies were predominantly based on laboratory experiments in order to better quantify the flow conditions and particle/oil concentration. The complex nature of OMAs has resulted in several challenges to simulate/mimic their formation and fate in numerical models, especially when they interact with the cohesive particles present in natural deltaic aquatic systems.

The first laboratory experiment related to minerals and oil droplets was presented by Delvigne [53], which suggested that oil types, mineral types, turbulence energy, and water salinity were the main influence on the aggregation process. Subsequently, numerous OMAs laboratory studies have been conducted and most indicate that spilt oil bio-degradation can be enhanced by increasing oil-mineral flocculation in a natural water column [54, 55, 56]. Floch et al., [51] quantified the amount of oil incorporated into OMAs and suggested that the extent of OMA formation is not significantly different from that of seawater for salinity values as low as 1.5–0.15 (1/20 to 1/200 of pure seawater). Omotoso et al., [48] found that the degree of oil-mineral interactions should be dependent on the viscosity of the crude oil and the type of mineral present, because the oil droplet surface changes with dynamic viscosity parameter.

A series of shaker jar tests by Khelifa et al., [50] indicated that although droplet size and shape were not correlated to oil viscosity, the concentration of oil droplets present during the flocculation process became extremely sensitive to the oil viscosity, temperature, and asphaltenes-resins content (ARC). Cloutier et al., [57] conducted an annular flume experiment to determine critical shear stress removing oil from a surface by resuspension and the effect of suspended sediment concentration (SSC) on the oil erosion processes. They found that SSC at 200–250 mg/l was observed to give maximum erosion efficiency and is therefore suggested as the optimal concentration for erosion and elimination of heavy crude oil at a water temperature of 13°C.

A more recent laboratory study was conducted by Sun et al., [58] to investigate the kinetics of oil-sediment aggregates formation under various mixing intensities using marine sediments (standard reference material 1941b) and Arabian heavy crude oil. Sun et al., [58] noticed that the formation of oil suspended particles aggregates increased exponentially throughout the mixing duration and peaked during a 5-hour period. Optimum trapping effectiveness of oil rose from 24 to 47%, and the necessary shaking (i.e. physical agitation) time lessened from 4.5 to 1.2 hours as both the mixing level and sediment concentration intensified. The maximum oil-to-sediment ratio reached 240–680 mg of oil per gram of sediment during the aforementioned 5-hour period. Observations of revealed that during more quiescent conditions, the oil-sediment aggregates formed were predominantly of a solid type and single droplet aggregates. While higher turbulent agitation resulted in more multidroplet type aggregates.

Moreover, many OMA studies have been focused on oil-mineral aggregates formation and their appearance using microscopic and imaging analysis in laboratory. Stoffyn-Egli and Lee [49] used laboratory protocols and microscopic methods to detect and classify the OMA types. They identified three types of OMAs: droplet, solid, and flake aggregates, and these are controlled by mineral types, oil surface properties, viscosity, and oil/mineral ratios. Moreover, they also report clear evidence that turbulence greatly enhances the OMA formation. Delvigne [59] used natural sediment in the laboratory experiment to study the physical appearance of oil contaminated sediment. Through his quantitative investigations, it was concluded that the division of oil in the different phase is affected by the oil-sediment interaction process, oil type, and oil concentrations.

Delvigne’s [59] microscopic observations of oil-contaminated sediment samples lead to three possible phases of presence of oil: oil droplets, oil-coated sediment particles, and “oil patches.” Therefore, both Stoffyn-Engli and Delvigne found the existence of consistent OMA types through their laboratory studies, including ① mineral particles attaching around the surface of big oil droplet, ② small oil droplets coated by sediments, and ③ flake-shaped oil-mineral aggregates. Khelifa et al., [50] reported more detailed OMAs study by using an epi-fluorescence microscopy to analyze the shape, size distribution, and oil concentration within OMAs. These results also show these types of OMAs with further quantification on more specific oil components. Furthermore, O’Laughlin et al. [60] used a high-resolution imagery method to record settling OMAs to analyze the particle size, density, and settling velocity. Their investigation revealed the settling velocities of artificially formed OMAs to be on the order of 0.1 ∼ 0.4 mm·s⁻¹. The OMA formation increases with suspended sediment concentration (>50 mg·l⁻¹).

Following the path led by all the abundant laboratory experiments of OMAs for the last decades (mentioned previously), numerical models have been adapted to predict the oil-mineral aggregation. Fitzpatrick et al., [61] presented a simple algorithm for oil-particle aggregates with the algorithm representing oiling conditions in a natural river delta, after the formation of oil-particle aggregates within the water column. Three groups of oil-particle aggregates have been included in the model, in recognition that there are likely multiple sizes and densities of oil globules and oil-particle aggregates in the riverbed. The groups range from 2 mm single oil globule with a 10 μm silt coating to more complex aggregates with multiple smaller globules and diameters of 31 μm and 100 μm. Densities range from just greater than the density of freshwater for the large oil globule with silt coating (1.034 g·cm⁻³) to somewhat heavier and close to the density of organic matter for the oil-particle aggregates (1.511 g·cm⁻³). Settling velocities range from 0.2 to 20 mm·s⁻¹, depending on the amount of oil relative to the size of the aggregate. A major assumption is that the oil-particle aggregates stay intact (i.e., they do not break up or disaggregate), during model runs. The simplified transport algorithm for oil-particle aggregates was suggested to be a start for future modeling of oil-particle aggregates formation, transport, and deposition. In addition, Zhao et al., [62] presented a new conceptual-numerical model, A-DROP, to predict oil amount trapped in oil-particle aggregates.

A new conceptual formulation of oil-particle coagulation efficiency has been introduced to account for the effects of oil stabilization by particles, particle hydrophobicity, and oil-particle size ratio on oil-particle aggregates formation. A-DROP was able to closely reproduce the oil trapping efficiency reported in experimental studies. The model was then used to simulate the oil-particle aggregates formation in a typical nearshore environment. Modeling results indicate that the increase of particle concentration in the swash zone would speed up the oil-particle interaction process; but the oil amount trapped in oil-particle aggregates did not correspond to the increase of particle concentration. It is suggested that the developed A-DROP model could become an important tool in understanding the natural removal of oil and developing oil spill countermeasures, by means of oil-particle aggregation.

1.3 Justification for this study

From the extensive literature review presented in the previous related studies, we deduce that flocculation of the oil-mineral aggregates is one of the keys to improve our understanding of the fate of spilt oil in coastal and deltaic aquatic regions. Most OMA studies focus on: the formation mechanism, physical appearance, and their influencing factors. However, extremely few researchers have systematically investigated the OMA characteristics in terms of how they affect and influence their resultant settling dynamics during the flocculation process. This presents a knowledge gap in predicting the amount and fate of oil throughout the water column and extending all the way to the seafloor, particular in muddy deltaic regions. The study presented here aims to fill this knowledge gap to better understand the settling dynamics and environmental impacts from the oil sediment flocculation in coastal and deltaic ecosystems, with the expectation of improving numerical modeling to predict the spilt oil fate in a more accurate, efficient, and reliable way.

OMAs have been proven both in laboratory and field to generate easily under turbulent flow environments in natural aquatic systems (including deltas) where there is a presence of cohesive materials, including minerals and biological matters. Because of the complex structure of natural OMAs, laboratory tests can provide more details on how specific natural materials (e.g., clay mineral) influence OMA flocculation and settling dynamics. Besides the mixing devices (magnetic stirrer and reciprocal shaker) to generate the OMAs and high-resolution digital microscopy camera to observe the details of OMAs, we also utilize a Vectrino to evaluate the generated turbulence and a state-of-the-art novel LabSFLOC-2 video camera system to observe the OMAs’ characteristics and most importantly their settling velocities. Collectively, a more robust and systematical study of OMAs has been conducted.

The chapter is divided as follows. Section 2 describes the experimental methodology and instrumentation utilized, with the main results reported in Section 3. A comparison of the OMA characteristics and a discussion of the implications to oil sediment transport and flocculation modeling are presented in Section 4, followed by some key points and concluding remarks summarized in Section 5.

2. Technical approach

2.1 Experimental design and instrumentations

A series of magnetic stirring jar (Figure 1a) and reciprocal shaking jar (Figure 1b) experiments have been conducted at the Center for Applied Coastal Research, University of Delaware, USA. Pure Kaolin clay, Bentonite clay, Xanthan gum powder (a proxy of biological Extracellular polymeric substance), and raw Texas crude oil (Dynamic viscosity: 7.27 cP) with various proportions have been used to generate different types of mineral flocs and Oil-Mineral Aggregates (OMAs) samples, including pure kaolin flocs, pure bentonite flocs, mixed kaolin-bentonite flocs, oil-kaolin flocs, oil-bentonite flocs, oil-kaolin-bentonite flocs, and oil-kaolin-bentonite-EPS flocs (mineral clay: 100–200 mg/l; oil: 50 mg/l). Artificial seawater (salinity ≈ 35‰) has been made from mixing clean water and pure salt. Magnetic stirring speeds have been set up to 500 rpm (Device range: 0 ∼ 1000 rpm) while reciprocal shaking speeds were at 120 rpm, 160 rpm, and 200 rpm for comparing the turbulence influence on the OMAs generating in jar tests. Both stirring and shaking for each experimental run last up to 2 hours and settling down for overnight, and it has been believed being long enough to generate mature OMAs for further studies.

Figure 1.

Jar experimental setup: a) shows the magnetic stirrer and b) shows the reciprocal shaking jar. A Vectrino was mounted above to monitor and record the three-way flow velocity.

A Nortek produced Vectrino (https://www.nortekgroup.com/products/vectrino) (Profiling Acoustic Doppler Velocimeter, sampling frequency in this work used 100 Hz) was used in stir-jar test to measure three-way velocities and calculate the statistics of turbulence. The Vectrino is a high-resolution acoustic velocimeter used to measure 3D water velocity fluctuations within a very small sampling volume and at sample rates of up to 200 Hz generally. It has a 30 mm profiling data collecting zone starting from 4 cm below the mid of sensor probes. It works by sending out a short acoustic pulse from the transmit element. The Vectrino was mounted on the shelf above the magnetic stirrer, and the sensor probes were in the water column (see Figure 1). The time series data set was collected in the same conditions (up to 500 rpm in stirrer and 200 rpm in shaker, artificial seawater and 110 mm diameter jar) with all the stir jar experiments with oil-mineral mixtures. Therefore, the turbulence of the mixing water column can be evaluated and provided as a constant value/pattern.

The mass settling dynamics of OMAs were observed using the low intrusive LabSFLOC-2 system (the second version of Laboratory Spectral Flocculation Characteristics instrument; [63, 64]) (Figure 2). This instrument was originally developed by Manning [65] originally, and it measures an entire floc population for each sample being assessed. LabSFLOC-2 utilizes a low-intrusive 2.0 MP Grasshopper monochrome digital video camera with a Sill TZM 1560 Telecentric (F4, with a magnification of 0.66 or 1:1.5) lens to optically observe all individual settling flocs (e.g., [66]) nominally 0.075 m above the base of a Perspex settling column (0.1 m square by 0.35 m high) within a 1 mm depth of field in the center of the column. Each pixel was 5 μm, with a 0.6% maximum distortion. More details can be found in [67].

Figure 2.

The LabSFLOC-2 setup on the desk beside the stir jar system for real-time samplings.

A high-resolution digital microscope system (Figure 3) has been used to obtain the floc detailed structures and number statistical analysis. All the floc samples were directly collected from the running experiment in real time and using wide mouth (> 2 mm) plastic pipettes (least original floc disturbance) transferring from the mixing jar to the microscope slides and then observing under 10-times zoom-in screen on a DELL laptop via the camera software provided by AmScope.

Figure 3.

High-resolution digital camera microscope.

A Beckman Coulter LS 13320 Particle Size Analyzer (Figure 4) located in the Advanced Materials Characteristic Lab, University of Delaware, can be used to determine the clay mineral components and particle size distributions of the floc or sediment samples. This LS 13320 MW high-resolution device observes individual particles (across the entire size range) using laser diffraction. The resultant sizes are accurately determined using the principles of Mie and Fraunhofer theories of light scattering.

Figure 4.

Beckman coulter LS 13320 particle size analyzer.

Considering the potential influences of oil and mineral clay addition to the flocculation initiation and floc structures, both adding oil before and after mineral clay into artificial seawater in the mixing jar have been tested and flocs images have been compared in Section 3.

2.2 Data processing and analysis

2.2.1 LabSFLOC-2 camera floc data

As one of the most novel floc video instruments, the LabSFLOC-2 produces not only visible individual floc images, but can also enable the estimation of essential quantitative floc properties (including floc size, effective density, settling velocities, etc.) through postprocessing of the raw floc image data. The recorded AVI files of streamed floc settling videos are not Codec compressed, so they can be analyzed with Matlab software routines. During postprocessing, the HR Wallingford Ltd. DigiFloc software–version 1.0 [68] is then used to semiautomatically process the digital recording image stack to obtain floc size and settling velocity spectra. A modified version of Stokes Law [69] then permits an accurate calculation of individual floc effective density [70].

Additionally, LabSFLOC-2 system provides the following supplementary individual floc information: floc porosity, floc mass, fractal dimension, floc shape, and mass settling flux. Manning et al. [64, 71] provides further details of both the LabSFLOC-2 floc acquisition procedures and postprocessing computations, respectively. LabSFLOC-2 provides data that covers many important aspects of flocculation; these floc data are necessities for comprehensively assessing and characterizing Oil-Mineral-Microbial settling dynamics and for improving the parameterization and calibration of numerical models.

In order to investigate the general spatial variation in the floc properties, a selection of the sample mean and Macrofloc:microfloc (using mean floc size 160 μm demarcation) parameterized floc properties have been summarized at the nominal LabSFLOC-2 acquisition height. A complete summary of mean floc parameters for all the samples in this report has been listed in Table 1, including floc size (D__Mean), mean effective density, mean settling velocity (Ws__Mean), and fractal dimensions (f_n).

Samples	Kaolin			Bentonite			Kaolin-Bentonite
Demacation	Mass	microfloc	Macrofloc	Mass	microfloc	Macrofloc	Mass	Microfloc	Macrofloc
Floe number	2631	2128	503	1705	681	1024	2998	2420	578
Mean floe size (μm)	120	101	199	185	105	238	104	76	225
Mean effective density (kg·m^−3)	315	336	224	127	187	87	410	461	200
Mean settling velocity (mm·s^−1)	2.41	1.82	4.89	2.00	1.08	2.61	1.97	1.27	4.90
Fractal dimension (nf)	2.54	2.54	2.55	2.30	2.30	2.30	2.57	2.59	2.53
Samples	Oil-Kaolin			Oil-Bentonite			Oil-Kaolin-Bentonite
Demacation	Mass	microfloc	Macrofloc	Mass	microfloc	Macrofloc	Mass	Microfloc	Macrofloc
Floe number	3102	2696	406	1592	580	1012	2610	1975	635
Mean floc size (μm)	102	86	204	198	115	246	120	87	222
Mean effective density (kg·m^−3)	249	269	113	127	167	104	408	446	290
Mean settling velocity (mm·s^−1)	1.21	1.03	2.41	2.53	1.07	3.36	3.33	1.94	7.63
Fractal dimension (nf)	2.40	2.40	2.36	2.33	2.29	2.35	2.57	2.57	2.58

Table 1.

Summary for microfloc and macrofloc of various types of oil-mineral flocs.

Floc sizes were measured from the image by overlaying an ellipse on each floc, which yields both major and minor axes of a given floc: D_x and D_y. This provides an indication of the floc shape in terms of the height/width aspect ratio. Settling velocity is determined by measuring the vertical distance from the center of each floc travels between two frames; the time step size between the two frames is known, which means that the floc settling velocity can be calculated by the ratio of distance to the time step size. To aid in the interpretation of the floc size data, the two orthogonal dimensions were converted into a spherical equivalent floc diameter, D_n, using equation:

$$D_n={\left(D_x.D_y\right)}^{0.5}$$E1

As mentioned, a modified version of Stokes Law [69] permits an accurate estimate of floc effective density:

$${\rho }_e=\left({\rho }_f-{\rho }_w\right)=\frac{W_{s}18\mu }{{\mathit{Dn}}^{2}g}$$E2

where g is the acceleration due to gravitational, and μ is dynamic molecular viscosity.

International Equation of State of Sea Water, 1980 [72] was employed to establish the relevant water density from measurements of water temperature and salinity. Floc effective density is the difference between ambient water density (ρ_w) and the bulk density floc (ρ_f). Eq. (2) enabled simultaneously measured individual floc settling velocities (W_s) and D to be directly related to the corresponding individual floc effective density.

To assist in data interpretation, the general trend of each floc population could be depicted by the calculations of various sample average floc parameters. The bimodality of each floc population [73] could be investigated by calculating the Macrofloc and smaller microfloc fractions [74, 75] as D_n > 160 μm and D_n < 160 μm, respectively [74].

2.2.1.1 Examples of LabSFLOC-2 floc size vs. settling velocity distributions

The previous section provided an overview of how floc properties can be obtained through LabSFLOC-2 measurements and subsequent postprocessing analysis on the floc images. We now illustrate in more detail the typical trend of individual floc sizes vs. flocs settling velocities. The scatterplots in Figure 5 depict individual spherical-equivalent dry mass weighted floc sizes (x-axis) plotted against their corresponding settling velocities (y-axis) of each experimental sample collected and analyzed by LabSFLOC-2 camera system.

Figure 5.

Plots of floc sizes vs. settling velocities of each type of mineral clay and oil mixtures. The three diagonal lines represent contours of Stokes-equivalent constant effective density (i.e., floc bulk density minus water density): Pink = 1600 kg·m⁻³ (equivalent to a quartz particle), green = 160 kg·m⁻³, and red = 16 kg·m⁻³. Solid black vertical line shows the separation of microflocs (<160 microns) and macroflocs (>160 microns); gray vertical line shows the average oil droplets size (57 microns).

Figure 5a shows the resultant flocs from a mixed Kaolinite and Bentonite suspension having floc size range from 15 to 508 microns. The plot shows a group of small-sized flocs (<100 microns) with low settling velocities (<0.2 mm/s) formed; based on comparisons, this fraction is lacking from floc populations composed of both pure Kaolin and pure Bentonite clays. Kaolinite-Bentonite flocs consist of a significant portion of small microflocs (81% of the 2998 flocs). Although the settling velocities range from 0.1 to 11 mm/s, the higher portion of microflocs causes a net reduction of averaged settling velocity to ∼2 mm/s.

When oil is added to Kaolinite-Bentonite mixture (Figure 5b), although the maximum floc size is 90 microns smaller than the corresponding non-oil suspension (Figure 5a), we observe more high-density Macroflocs with settling velocities greater than 10 mm/s (peaking at 20 mm/s). We also observe 5% more Macroflocs (by population), but their effective density is quite widely spread; from very porous Macroflocs with a density close to water (i.e., below the red line), to those Macroflocs of a higher density—approaching the density of a quartz particle flocs (i.e., the pink line).

2.2.2 Vectrino data (turbulence)

The Vectrino data recorded 5–10 minutes continuously three-way velocities in the mid of jar water column as an interval of 30 mm from the 4 cm below its transducers. The raw data file.DAT has been input into Matlab for statistical analysis. Three approaches [71, 76, 77] have been utilized to calculate the turbulence kinetic energy (TKE), dissipation rate (ε), and the flow shear parameter G and for comparing the reliability and accuracy (Figure 6).

Figure 6.

Turbulence kinetic energy (TKE), dissipation rate (ε), and the flow shear parameter G calculated by approaches from Khelifa [71, 76, 77], respectively.

In Khelifa’s [77] study, a constant l_t (≈ 47 mm) has been assumed, which represents characteristic velocity and length scales of turbulence in a specific jar. Therefore, turbulence dissipation rate (ε) has been calculated by $\frac{1}{0.047}{\left(\frac{2}{3}\mathit{TKE}\right)}^{\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}$. On the other hand, Manning et al. [71] used the shear parameter G (units of s⁻¹), which is the root-mean-square of the gradient in the turbulent velocity fluctuations, and turbulent shear stress τ to evaluate the turbulence [78]. These are related through the shear velocity u* by $G={\left(\frac{u^{\ast }3\xi }{\mathit{\kappa \nu z}}\right)}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}$, where κ is the von Karman’s constant (taken as 0.40), ν is kinematic viscosity of the water, z is height above the bed, h is water depth, and ξ = 1 − z/h. Different from the dependence on specific conditions of the former two approaches, Huang [76] calculated dissipation rate (ε) by common frequency spectrum (φ(f)) using Taylor’s frozen turbulence hypothesis [79] to give: $\phi \left(f\right)={\left(\frac{U}{2\pi }\right)}^{2/3}\alpha {\epsilon }^{2/3}{f}^{-5/3}$, where f is the frequency, and U is the mean velocity. Thus, εcan be calculated by $\frac{2\pi }{U}{\alpha }^{-3/2}<f^{5/2}{\phi }^{3/2}\left(f\right)>$.

2.2.3 Microscope images analysis

The images collected from the digital microscope of each floc sample (e.g., Figure 7) have been number counted and shape analyzed individually for further statistical analysis, and particles flocculation stickiness rate discussions. The temporal development of flocs has been analyzed by using the evolutionary changes in the floc numbers observed during the microscope experiments; this technique was also utilized to represent the stickiness of flocs in time series (see [80]). For each mineral clay flocculation experimental run of the jar tests, floc samples have been collected in real time and screen observed under microscope at the time interval of 1 minute, 2 minutes, 4 minutes, 8 minutes, 12 minutes… 140 minutes. Then all the individual floc numbers have been counted manually for each microscope image (e.g., Figure 8). The numbers of flocs formed in temporal indicate the development of the flocculation, which has been used to represent the floc particles stickiness rate parameter (result will be presented in Section 3.3).

Figure 7.

Example of oil-mineral aggregates images viewed under high-resolution microscopy.

Figure 8.

Floc number manually counted from the microscope images. The red part showing the individual floc can be recognized manually for further statistical analysis.

Additionally, averaged oil droplets size has been analyzed by averaging the oil droplet size from the microscope images (e.g., Figure 9). Delvigne et al., [53, 81, 82] report that the level turbulent activity and the viscosity of oil both heavily control oil droplet sizes in turbulent flows. Observations of the distributions of dispersed oil droplets sizes (D) were seen to follow a pattern corresponding to N ∼ D^−2.3; N is the number concentration of oil droplets of size D. This empirical interrelation was valid for oil dispersion due to breaking waves during the short period plunging phase [82]. Furthermore, pouring experiments conducted by Delvigne and Hulsen [83] showed that the oil dispersion coefficient was not affected by oil viscosity when the oil viscosity was <1 cm ²/s.

Figure 9.

Pure oil droplets samples and image from microscope for manually counting the averaged droplet size.

In contrast, at higher oil viscosities, the dispersion decreases considerably with viscosity. Fraser and Wicks [84] have discussed how the concept of critical Weber number [85] may be used to estimate the maximum size of stable oil droplets dispersed in sea. Li and Garrett [86] have investigated theoretically the effect of oil viscosity on the size of droplets, assuming that viscous shear is the mechanism for droplet breakup. Their research showed that the ratio (μ_d/μ_c)ⁿ was comparable with the largest droplet present, whereby μ_c is the viscosity of the uninterrupted (i.e., continuous) phase, μ_d is the viscosity of the oil droplet. The superscript term “n” is equivalent to 3/8 when the oil droplet size exceeds 50% of the Kolmogorov length scale [87], or otherwise it reverts to 1/8.

3. Results

3.1 Oil droplet size

The pure oil droplet sizes have been statistically analyzed from 10 selected microscope images. The results show the median size of the mass oil droplets is 60 microns, mean size is 57 microns, minimum size is 12 microns, maximum size is 120 microns, and 90% of the entire oil droplets observed are <80 microns, and 99% are <110 microns. Compared with the previous oil droplets size studies mostly for the dispersed oil droplets size, these are slightly larger. Lunel [88] also estimated that 90% of the oil droplets (50% of the oil volume) were < 45 μm in diameter, and that 99% of the oil droplets (80 ∼ 90% of the oil volume) were < 70 μm. In addition, Khelifa et al. [50] suggest that considering the mechanism of droplet breakup is controlled by oil and continuous phase properties and the energy dissipation rate εby turbulence.

3.2 Floc structure

Both Kaolin and Bentonite cases samples have been analyzed using a microscope, and the high-resolution (10-times zoomed in scale) images provided the details of the floc structure for each mixture. Three basic floc types have been identified: i) mineral flocs/aggregates, ii) oil droplets attaching/combining Kaolin aggregates, and iii) stringy/flake-shaped oil-Bentonite aggregates.

3.2.1 Kaolin floc and oil-kaolin aggregates structures

Figure 10a shows a representative sample of settled pure kaolin clay floc (Kaolin clay: 100 mg/L), which is about 250 microns in length. The floc was generated under a 2-hour constant turbulent condition (turbulence dissipation rate: 0.015 ∼ 0.02 m²/s²). With the addition of 100 mg/L Texas crude oil in the mixing jar (oil added into mineral floc mixtures), the oil droplets can be observed being attached or half embraced within the kaolin clay structures (such as Figure 10b and c). Basically, we observed mineral flocs of Kaolin clay attached or combined with the oil droplets (surface) and developed lager oil-Kaolin aggregates. This is consistent with one of the OMA types shown in the previous studies (such as [49, 50]). They found that droplet OMA (or oil coated by sediment aggregates) appears as oil (spheres) with mineral particles or flocs attached to their surface only. In this case, the droplets do not enclose mineral particles. The quantity of minerals attached to a droplet is highly variable.

Figure 10.

Floc images from the high-resolution digital microscope camera. a) the pure kaolin clay floc structure; b) & c) oil and kaolin clay mixed floc structure; d) the pure bentonite clay formed flocs structure and e) & f) oil and bentonite clay formed flocs, & g) mixed OMAs with oil-kaolin-bentonite flocs.

3.2.2 Bentonite floc and oil-bentonite aggregate structures

The Bentonite clay cases (Bentonite clay: 100 mg/L), generated with same turbulent dissipation rate, are shown in Figure 10d–f. Pure Bentonite clay particles can be much more attachable than the Kaolin clay particles (more quantitative comparison on flocculation rate will be given later). The Bentonite flocs are long stringy-shaped and up to 100 ∼ 200 μm in width and several hundred microns in length (Figure 10d). However, largely different from the oil-Kaolin case, the oil-Bentonite mixture flocs show an entire reshaped oil structure. The sphere-shaped oil droplets have now disappeared and formed more randomly stringy/flake-shaped oil-Bentonite aggregates, up to hundreds of microns in size as shown in Figure 10e and f.

Compared with the previous studies, the oil-Bentonite aggregates observed here belong to the category of flake OMA. Flake aggregates have the appearance of membrane structures, usually floating or neutrally buoyant, which can attain hundreds of microns in length. They have been observed in several laboratory studies (such as [49, 50]). From inspection of the microstructure of oil-Bentonite aggregates, they appear extremely orderly with a “feather-like” or “dendritic” appearance. These flakey structured aggregates tend to collapse under fast or extended agitation during high shear stress episodes. The resultant crumpled flake aggregates are less buoyant and more compact and often depicted as a solid OMA. However, they can be distinguished from solid OMA, by their favorable mineral pattern organization [49].

3.2.3 Mixed oil-kaolinite-bentonite aggregate structures

After mixing equal amounts of Kaolinite and Bentonite clay together with oil in the mixture, the resultant comprises both droplet OMAs and flake OMAs present in the same floc population (Figure 10g). From inspection, the large flake-shaped OMA observed in Figure 10g has similar floc sizes to those in the oil-Bentonite case (see Figure 10e–f). In the next sections, we will investigate different mineral flocs and OMAs settling velocities and discuss their relationship to floc structures.

3.3 Floc characteristics

3.3.1 Mean floc size, macroflocs and microflocs distributions

In terms of the results from the jar tests, the comparisons between the results in Table 1 reveal the oil, kaolinite, and bentonite components’ influence on the OMA floc characteristics, respectively. A systematic comparison is given next.

3.3.1.1 Floc characteristics between pure kaolin, bentonite flocs, and mixed kaolin-bentonite flocs

A comparison between Kaolinite and Bentonite flocs shows that the Kaolinite flocs have around 50% higher floc number and 50% smaller floc size in total. Interestingly, the larger number of flocs in Kaolinite is only due to microflocs. Simply in terms of Macrofloc number, Kaolinite has only half of that of Bentonite. Overall, the Kaolinite floc effective density is around 2.5 times higher than that of Bentonite flocs, which results in approximately 20% quicker settling velocity of Kaolinite flocs than that of Bentonite flocs. Finally, the fractal dimension of Kaolinite flocs is about 2.4, which is higher than that of Bentonite flocs of around 2.0. With these floc characteristics, we can conclude that the Kaolinite clay has lower flocculation rate than that of Bentonite, and this also suggests that Kaolinite clay shows a lower relative cohesivity than Bentonite clay.

After mixing equal amounts of Kaolinite and Bentonite clay (50% and 50%), the resultant mixed floc sample shows slightly (14%) higher floc numbers in total than that of Kaolinite floc sample and significantly more flocs (1.75 times) than the total present in the corresponding Bentonite floc sample. Meanwhile, the mixed floc sample has the smallest averaged floc size (104 microns) and settling velocity (1.97 mm/s), but the largest averaged total effective density (410 kg/m⁻³) when comparing with those of the two pure clay samples.

Noticeably the Macroflocs in the mixed Kaolinite-Bentonite sample shows medium averaged floc size and averaged effective density of 225 microns and 200 kg/m⁻³, respectively; that places the mixed Kaolinite-Bentonite sample midway between pure Kaolinite (199 microns, 224 kg/m⁻³) and the pure Bentonite (238 microns, 87 kg/m⁻³) cases. And the averaged Macrofloc settling velocity (4.9 mm/s) of mixed Kaolinite-Bentonite is similar to that of pure Kaolinite (4.89 mm/s), but much faster than that of pure Bentonite (2.61 mm/s). This indicates a signal that within the mixed clay case, the Macroflocs development is somewhat enhanced by the more cohesive bentonite component, but the entire mixed floc characteristics are still dominated by the effect of the less cohesive Kaolinite, especially via the microfloc population. This seemingly subtle point is raised here because it may play a more important role when interacting with oil droplets (see the next section).

3.3.1.2 Floc characteristics between pure kaolin, bentonite clay flocs and oil-kaolin, oil-bentonite flocs, respectively

The addition of a crude oil component into the Kaolin floc sample, oil-Kaolin floc numbers increased by around 18% in total, while the total averaged floc size decreased by around 18%, and the average total effective density also decreases by 20%. As a result, we obtain a reduction of averaged total settling velocity by half. Therefore, the oil addition decreases the Kaolinite flocculation rate (i.e., it produces smaller flocs) and the lower density of oil droplets further reduces the OMA density. Noticeably, for Kaolinite, its OMAs show a significant reduction in settling velocity. This is particularly demonstrated in the Macrofloc fraction: adding oil reduces the number of Macroflocs, and increases mean Macrofloc size and fractal dimension very slightly. It is the significant reduction of mean effective density (reduce by half) that causes a twofold slowing of the Macrofloc OMAs settling velocity.

For the Bentonite case, adding an oil component decreases the total number of flocs by around 7%, while the mean total floc size increases by around 7%. Although the averaged total effective density is in the same level, the averaged total settling velocity increases by 25%, which may be caused by the increase of floc size. In Macroflocs, although the mean floc number, size. and fractal dimension stay in same levels, their mean effective density and settling velocity tend to be higher after added oil. Therefore, for the entire floc group, adding oil to the matrix slightly increases flocculation in the oil-Bentonite case, and for its Macroflocs, effective density and settling velocity both rose by adding oil. Here, we can see that the response of Kaolinite and Bentonite to the addition of oil is distinctly different, and this is related to the very different OMA structures illustrated in Figure 10.

3.3.1.3 Floc characteristics between mixed kaolin-bentonite flocs and oil-kaolin-bentonite flocs

In total, floc number shows a reduction of around 18% by adding oil into a mixed Kaolinite and Bentonite suspension, and the averaged total floc size contrarily increases by around 15%, which may indicate an increase of floc cohesivity due to the addition of oil droplets. However, a more careful observation suggests that the changes due to the addition of oil show a reversed trend between the microfloc and Macrofloc fractions. Namely, by adding oil the microfloc flocculation rate is increased, which causes an increase in the microfloc mean size. In contrast, by adding oil to the suspension, the Macrofloc flocculation rate decreases and translates to a net rise in floc density. Significantly, by adding oil, the mean settling velocity quickens by almost 70%; this is due to the increase in the resultant average microfloc size, plus a marked rise in floc density within Macroflocs.

Overall, adding oil to the mixed Kaolin-Bentonite suspension reveals a more unexpected response. Our results show that in terms of the floc numbers, size, density, and settling velocity, the addition of oil to a Kaolin-Bentonite mixture suggests a slightly more exacerbated change in floc property patterns, when compared with the same floc characteristics from adding oil into pure Bentonite flocs. Therefore we deduce, in Kaolin-Bentonite mixtures, oil is selectively actively interacting noticeably more with Bentonite rather than with Kaolinite clay minerals (under similar clay content conditions), respectively.

3.3.2 Size classes analysis

3.3.2.1 Settling velocity

The averaged settling velocity of each size class is shown in Figure 11. For the Kaolin case, the pure mineral flocs show a marked quickening of the settling velocities as the floc sizes grew through the 32–512 micron subfractions. By adding oil to Kaolinite, although the settling velocities initially still show the same increase through the 32–256 microns flocs, on reaching the 512 micron class, the settling velocity abruptly reduced by half.

Figure 11.

Averaged settling velocity of each size classes flocs. a) the kaolin and oil-kaolin cases, and b) the bentonite and oil-bentonite cases and c) mixed kaolin-bentonite and oil-kaolin-bentonite cases. The numbers on the top of the bars show the floc numbers in each class.

Bentonite clay suspensions depict a very different flocculation trend (when compared with the corresponding Kaolinite flocs) for both pure mineral and oil-Bentonite cases. Bentonite produces a rise in settling rates with the growing flocs, especially in the larger size fractions from 512 to 1024 microns. The settling velocity doubles by adding oil into Bentonite flocs.

Combined Kaolin and Bentonite mixtures again show an increasing settling velocity with growing floc sizes trend for both for the mixed minerals flocs and oil-Kaolin-Bentonite flocs. The dramatic rise in of settling velocity by adding oil occurs in the 512-size class; the 1024 class present for Bentonite-only flocs has now disappeared.

3.3.2.2 Effective density

For a clearer idea of floc effective density trends, seven size groups have again been utilized as shown in Figure 12. Both Kaolin and Bentonite flocs show the decreasing propensity of the effective density through increasing size classes, and a similar trend once oil is added. Effective density decreases at a quicker rate for pure Bentonite when compared with pure Kaolinite (Figure 12b). However, the addition of oil into these both types of mineral clay flocs saw changes predominantly in size class 256–512 microns for Kaolin case and 32–64 microns for Bentonite case, where the oil concentrations are most probably more abundant. More specifically, adding an oil component tended to reduce the floc effective densities, especially dramatically in the 256–512 micron class; the main exception being the 16–32 micron class, which shows slight increases in effective densities (Figure 12a).

Figure 12.

Averaged effective density of each size classes flocs. a) the kaolin and oil-kaolin cases, b) the bentonite and oil-bentonite cases and c) mixed kaolin-bentonite and oil-kaolin-bentonite cases. The numbers on the top of the bars show the floc numbers in each class.

On the other hand, an oil addition to the sedimentary matrix instigated an increase in the effective densities in oil-Bentonite clay flocs, especially in the 32–64 micron class; the only exception being the 64–128 micron class. Therefore, the most noticeable density decrease was due to the addition of oil to Kaolinite within the Macrofloc fraction (256- and 512-micron classes). The most significant increase of density when oil is added to Bentonite is at the microfloc class 32–64 micron, as well as Macroflocs of 1024 micron. Noticeably, at 16–32 micron class, there may well be no oil-floc (for both Kaolinite and bentonite) due to oil droplet sizes.

3.3.2.3 Fractal dimension

The fractal dimension changes in size classes of both Kaolin and Bentonite cases also reveal that the oil addition decreased fn largely in the 256–512 microns floc size range for Kaolin flocs, and 32–64 microns for Bentonite cases (Figure 13). On average, Kaolinite flocs have larger fractal dimensions (fn between 2.4 and 2.6) than those of Bentonite flocs (fn from 2.2 to 2.3). The addition of oil decreased the fractal dimensions in Kaolinite case, particularly showing a substantial reduction in the 256–512 micron class, with the exception of the 16–32 micron class where it rose slightly.

Figure 13.

Averaged fractal dimensions of each size classes flocs. a) the kaolin and oil-kaolin cases, b) the bentonite and oil-bentonite cases. The numbers on the top of the bars show the floc numbers in each class.

In contrast, the oil-Bentonite flocs exhibit larger fractal dimensions than corresponding Bentonite flocs particularly in the 32–64 micron range the exception being the 64–128 micron class. Therefore, the consequence of fractal dimension changes after adding oil is primarily in the Macroflocs (256–512 micron) for the Kaolinite case, and more so in smaller microflocs (32–64 micron) for the Bentonite case.

3.3.2.4 Mineral flocculation stickiness rate development as a time series

Flocculation occurs and trends to be stable in mineral clay particles with the mixing time under a constant turbulent condition. With the flocculation development, smaller mineral aggregates can stick together becoming newer lager flocs, which leads to a number decay with the same total primary particles condition. Figure 14 shows the Kaolin (orange solid circles) and Bentonite (blue solid circles) clay particles flocs were developing with the decreasing of total floc numbers. From the first 10 minutes, the floc numbers reduced dramatically in both Bentonite and Kaolinite flocs. Between 10 and 30 minutes, the floc numbers demonstrate much more stability, but still a low level in fluctuation. After 1 hour, the trend in floc numbers seems to have become more stable.

Figure 14.

Statistical result of the temporal development of both bentonite and kaolinite floc numbers, derived from the averaged size of mass flocs from microscope images analysis.

It is considered that this widespread decay rate in floc numbers can be attributed to the mineral flocculation stickiness rates between each temporal scale. Both the Kaolin and Bentonite floc development showed similarity, although the Kaolin floc numbers tended to reduce dramatically within the first 10 minutes, there was much less fluctuation in the floc numbers after the 20-minute period. The floc numbers keep stabilized from around 30 minutes and until to the end of the test.

Overall, the floc number decay rate, which can represent the stickiness of mineral clay flocs, occurred more rapidly for Bentonite clay floc, and was much more stable than that of Kaolinite flocs. It is recommended that a higher frequency of temporal data is required for further detailed analysis, especially for the initial 10-minute phase.

4. Discussions

The laboratory experimental results presented above provide sound information of the multitypes of pure mineral clay, oil-mineral flocs characteristics, and their respective flocculation dynamics under a certain constant flow turbulent condition and seawater environment; these can be representative of the deltaic environments (including coastal and estuarine). The two clay mineral types, Kaolin and Bentonite, formed into distinctly different OMAs. Their implication for future contributions to oil-contaminated sediment transport modeling is discussed in the following sections.

4.1 Comparison of oil droplets, oil-kaolin and oil-bentonite aggregates

The oil droplets formed under the turbulent mixing in the experimental jar tests resulted in a mean size of 57 microns, with a minimum and maximum droplet size of 12 and 120 microns, respectively. 90% of the entire oil droplets observed were < 80 microns, and 99% were < 110 microns. The induced turbulence (ε ≈ 0.015 m²/s³ and G ≈ 120 s⁻¹, Figure 6) was characteristic of a highly turbulent region, when compared with modeling predictions of the estuarine flow conditions [89].

Under the consistent experimental conditions and concentrations, microscopy analysis reveal for the oil-Kaolin aggregates, the clay mineral particles/flocs adhere on the oil droplets surface, and the mineral particles act as a web structures surrounding the oil droplet, thus preventing it attaching to other oil droplets or further rebonding to oil slicks. Since the Kaolin mineral particles can be attached together as a much larger structure than an individual oil droplet, the oil can be observed being attached or even embraced within the Kaolin flocs (Figure 10a–c). And the oil droplet can be safely preserved with the Kaolin clay flocs after the entire oil droplet surface area has been occupied by mineral particles. This type of oil-mineral aggregate has been proven to be the same with previous OMAs studies, such as Zhao et al. [62].

On the other hand, with the equivalent concentration of oil and mineral clay particles and under the same mixing turbulent conditions, Bentonite clay particles tend to reshape the oil droplets into large (∼ as large as 900 microns floc has been observed) stringy oil-Bentonite aggregates. From the beginning, the Bentonite particles form low density and large size predominantly stringy-configuration mineral flocs. With the addition of oil droplets, the Bentonite mineral flocs start attaching and combining with the oil. After a number of hours of mixing and flocculating, the oil component can be absorbed in the high porosity Bentonite flocs and adopt the characteristics we observe of oil-Bentonite aggregates (Figure 10d–e).

In terms of mixed kaolin-bentonite flocs, the results indicate a preference for kaolinite-like or bentonite-like properties within different size classes. Ye et al. [41] noted pronounced kaolinite-like features in the smaller microflocs, whereas the larger macrofloc size fraction produced bentonite-like features. These are important factors when considering the mineral interaction with oil droplets.

A conceptual structure of both oil-Kaolin and oil-Bentonite aggregates are illustrated in Figure 15a and b, respectively. Previous research has revealed that the quantity of oil droplets stabilized by oil particle aggregates formation can be strongly influenced by the ratio of oil to number of particles, plus the individual aggregate sizes and shapes [90, 91, 92]. But more significantly, the original cohesive sediment type has been proven as a key of oil-mineral aggregates structure, which can directly influence all other physical characteristics of the OMAs.

Figure 15.

The illustration showing the conceptual structure of kaolin clay formed oil aggregates and bentonite clay formed oil aggregates with difference.

4.2 Implications to the oil sediment transport and flocculation modelings

The oil-Kaolin flocculation has been proven to be mineral clay particles/flocs attaching or embracing original oil droplet together until the oil surface area fully occupied by clay particles to develop new OMAs and settling or transport through the water column. From the mass population statistical analysis, floc numbers slightly increased after adding the oil into the Kaolin mixtures because of the contribution of oil droplets to the floc entire numbers. Although the averaged floc size and effective density slightly reduced by adding oil, the average settling velocity reduced by half with the oil addition, from 2.4 mm/s in pure kaolin flocs to 1.2 mm/s with oil-Kaolin flocs. This indicates that the oil addition has a low-level influence on the Kaolin floc size and numbers, but clearly decreases the effective density and settling velocity in the mass populations. However, the plots cluster overlap area highlighted in Figure 16 also reveals that the main difference between pure Kaolin flocs and oil-Kaolin flocs is located in the upright area of the population cluster plots where the oil-Kaolin floc settling velocities are unchanged, whilst the pure Kaolin flocs depict a trend of quickening settling velocities with the growing floc size. Clearly, the difference between the oil contaminated flocs and pure mineral flocs is within the macrofloc fraction (floc size >160 microns).

Figure 16.

Plots of floc sizes vs. settling velocities of kaolin flocs (light blue) and oil-kaolin mixing flocs (red). The information in the boxes shows the mass population and MACROfloc subpopulation statistically results, respectively. The three diagonal lines represent contours of Stokes-equivalent constant effective density (i.e., floc bulk density minus water density): Pink = 1600 kg·m⁻³ (equivalent to a quartz particle), green = 160 kg·m⁻³, and red = 16 kg·m⁻³. Solid blue vertical line shows the separation of microflocs (<160 microns) and macroflocs (>160 microns).

Therefore, it can be accepted that taking the macrofloc subpopulation analysis into consideration is key to a clearer understanding of oil-Kaolin flocculation, and can be more representative, even though the macrofloc fraction comprises less than 20% of the total population mass. From the statistical results of the macrofloc subpopulation, the reduction in both their effective density and settling velocity is larger than those in the entire population. This suggests that the oil addition is most likely influencing these particular floc characteristics more sensitively, and it is more accurate to compare with pure mineral flocs and oil-mineral flocs within this subpopulation. These factors can be used to enlighten the present oil sediment and flocculation modeling work, essentially because the present oil-particle aggregates models are primarily based on the assumptions of the structures of oil-Kaolin flocculation process (such as, [61, 62]).

Conversely, for mass populations, the oil-Bentonite flocs show almost no difference with Bentonite flocs statistically in terms of their floc numbers, effective density, settling velocity, and porosity (Figure 17). However, the macroflocs show a nominal 30% rise in effective density and settling velocity in Oil-Bentonite flocs, when compared with pure Bentonite flocs; this reveals an opposing situation to that of the Kaolin case.

Figure 17.

Plots of floc sizes vs. settling velocities of bentonite flocs (light blue) and oil-bentonite mixing flocs (red). The information in the boxes shows the mass population and macrofloc subpopulation statistically results, respectively. The three diagonal lines represent contours of Stokes-equivalent constant effective density (i.e., floc bulk density minus water density): Pink = 1600 kg·m⁻³ (equivalent to a quartz particle), green = 160 kg·m⁻³, and red = 16 kg·m⁻³. Solid blue vertical line shows the separation of microflocs (<160 microns) and macroflocs (>160 microns).

This can be explained by the unique structure of oil-Bentonite aggregates, which radically differ from oil-Kaolin aggregates (as discussed in an earlier section). Rather than the OMAs forming from simply attaching oil droplets and mineral particles/flocs (as with Kaolin), the Bentonite flocs and oil droplets become mutually absorbed. Therefore, the oil-Bentonite aggregates tend to be more robust in structure and larger in size than their corresponding oil-Kaolin aggregates, which leads to an increase in both effective density and settling velocity for the Bentonite types. This more complex Bentonite flocculation structure can pose additional issues from the numerical modeling perspective in cohesive sediment flocculation. However, since the Bentonite clay is one of the most common mineral particles in natural environments, this tends to be a necessity for the future cohesive sediment transport and flocculation modeling improvement within muddy deltaic regions.

5. Summary of key points

• There is a vast difference in OMAs structures between Kaolin and Bentonite minerals.

• The difference between pure mineral clay flocs and OMAs characteristics reveals that the addition of oil instigates a net reduction in the floc size, effective density, and settling velocity of flocs formed from Kaolin. However, the presence of oil in the sedimentary matrix increased the effective density and settling velocity of flocs in Bentonite cases. This is because the Kaolin particles/flocs directly attach to oil droplets to form OMAs, while the Bentonite mineral flocs interact with the oil component more by absorbing each other to form more structurally complex oil flakes and strings.

• Macrofloc fractions of OMAs can be focused on to improve the OMA models, because the addition of oil has been proven more sensitive to macroflocs than the microfloc group.

• The inclusion and parameterization of OMA data for the use in numerical modeling (e.g., [93, 94]) require further development and continued refinement, particular for regional deltaic systems.

• OM-EPS Aggregates (e.g., [80]) need to be studied extensively to predict the natural oil contamination and environmental changes after oil spills. 3D assessment techniques such as those demonstrated by Zhang et al. [95] and Spencer et al. [96] may provide further enlightenment.

Acknowledgments

The research reported in this chapter was primarily part of the Consortium for Simulation of Oil-Microbial Interactions in the Ocean – CSOMIO that was funded by GoMRI - Gulf of Mexico Research Initiative (Grant no. SA18–10), together in part by the National Science Foundation (NSF) OCE-1924532. AJM’s contribution toward this chapter was also partly assisted by the TKI-MUSA project 11204950-000-ZKS-0002, HR Wallingford company research FineScale project (Grant no. ACK3013_62), and NSF grant OCE-1736668. LY’s contribution was also partly supported by Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) (No. 311020003). The datasets presented in this study can be found in online repositories GRIDDC and can be found at: https://data.gulfresearchinitiative.org.

Conflict of interest

The authors declare no conflict of interest.