Evaluation of Oil-Sediment Interactions and the Possibility of Oil Sinking in Marine Environments

With a continuous growth in global demand for crude oils as energy source, increased exports of oil sands products from Canada have been proposed by industry. To provide access to new markets, more transportation options are being considered including shipping by marine tankers from coastal ports to foreign markets. These marine transportation activities pose a potential risk for a dilbit spill at sea. Following a spill in the marine environments, the oil experiences a series of weathering processes which alter its physical and chemical properties and drive its behavior and fate in the environment. One major concern is whether spilled oil will float or sink after its exposure to the natural processes. Sunken oil is more difficult and expensive to find and cleanup and can contaminate sediments and presents a toxicity risk to benthic species (Dew et al., 2015).

The knowledge of properties of diluted bitumen products is crucial to understand and predict their fate and behavior if spilled in the environment. Diluted bitumen products have a density lower than freshwater, suggesting that the fresh dilbit will initially remain on the water surface until modified by weathering process such as evaporation and sediment uptake. (Crude Quality, 2015) Thus, from an oil spill response standpoint, it is desirable not only to have information about the fresh product as produced or in the pipeline, but also knowledge of the changes in its properties after the product was exposed to environmental weathering processes. The most important weathering process in the early stage of a spill is evaporation. Dilbit is produced from a mixture of low-viscosity diluent, typically condensate, with a heavy bituminous material. It has been widely supposed that the lighter fractions which compose the majority of the diluent would evaporate very quickly when dilbit is released. As these tend to be very volatile, in gasoline range, this has raised concerns that the physical properties of the product after spilling could change rapidly and significantly, creating the potential for sinking of the weathered product.

Evaporation affects various physical properties of dilbit. One of the most critical to environmental behavior and fate is density. The results of previous work (Government of Canada, 2013) showed that evaporation could increase dilbit densities to levels close to or even exceeding freshwater density, but not of saltwater, indicating that evaporation alone is unlikely to be sufficient to cause sinking of the dilbit products in marine environments.

Another process that could lead to oil submergence is the mixing of oil with suspended sediment particles in the water column. Knowledge and experience obtained from observations made following the 2010 Marshall Michigan spill in the Kalamazoo River indicated the uptake of suspended sediments by spilled dilbit played a key role in driving the sinking of oil (NTSB, 2010; U.S. EPA, 2013). A large number of studies have been conducted to investigate the oil-particulate interactions in varied aquatic environments (Khelifa et al., 2002; Khelifa et al., 2005; Loh et al., 2014; Gong et al., 2014; USGS, 2015; Hospital et al., 2016, O' Laughlin et al., 2016; Wu et al., 2016). However, there is still limited information on how dilbit behaves when released into marine environments; therefore there is a continued need to advance research to expand the knowledge of dilbit weathering, fate, and behavior. This study is an extension of our previous work on dilbit properties and spill behavior (Government of Canada, 2013). In previous work, we characterized the changes in important spill-related properties of dilbit with evaporative weathering, observed the fate of oil (i.e. sink or float) in salt water when exposed to high level of sediment, and qualitatively reported on the nature of the OPAs formed to better understand the nature of dilbit-sediment interaction.

In this study, the oil-sediment mixtures formed are father characterized. The same Cold Lake blend was selected for testing because it is one of the highest-volume products transported by pipeline in Canada. Both fresh and artificially weathered CLB were studied considering the fact that dilbit will rapidly weather once it is spilled. To provide greater comparability with the dilbit behavior simulations, a light conventional crude oil, an intermediate fuel oil, and a heavy fuel oil were also included in this study. As for sediment, in addition to the mineral sediments (kaolin and sand) used in the previous work (Government of Canada, 2013), a natural sediment from the Douglas channel in northern British Colombia was used to represent more realistic conditions. The particle size distribution of the OPAs was measured by laser diffraction technique. The morphological characteristics of the OPAs were observed by microscopy. Densities of the OPAs were evaluated through buoyancy test in a series of water and ethylene glycol (or glycerol if higher density fluid was desired) mixtures in various ratios.

Three different sediments were used for this study. Kaolin mineral fine sediment (i.e., hydrated aluminum silicate with 90% below 5.8 μm particle size) and sands (silicon dioxide, 200–400 μm particle size) were purchased from Sigma-Aldrich (St Louis, MO). A natural benthic sediment, referred to as KA04, (90% 2–63 μm particle size) was collected from Kitimat Arm in Douglas channel (BC, Canada). The diluted bitumen, Cold Lake Blend-Winter (abbreviated as CLB in this study), was chosen for the evaluation of physiochemical properties, weathering behavior and oil-sediment interaction in water in this study. The weathering process was performed following the optimized procedure published by Fieldhouse et al. (2010). CLB was studied at fresh (0% weathered), 15.75% (moderately weathered, W2) and 25.20% (heavily weathered, W4) fractions. For comparison, three other fresh oils at different viscosity levels were used as reference oils for this evaluation, including a standard light crude oil (Alberta sweet mixed blend/ASMB), an intermediate fuel oil (IFO 180), and a heavy fuel oil (Bunker C).

As with the previous study (Government of Canada, 2013), the major goal of this work was to characterize the possible types of OPA formed, rather than to examine the likelihood or dynamics of formation. Thus, as was the case previously (Government of Canada, 2013), a sediment loading of 10 g sediment per litre of water was used to ensure that the test media were not starved of sediment which might inhibit or limit any formation of aggregates or mixtures. Preliminary testing with 100 mg/L sediment concentration resulted in very low efficiency in formation of oil-particulate aggregates. Loadings of 10 g sediment/L are similar to levels of sediments found in the annual maxima of coastal river outflows. Uncles et al., showed that depth-averaged suspended sediment concentrations are from 1.0 to 48,400 mg/L in a set of 44 estuaries measured in the deltas at the site of maximum turbidity at high tides (Uncles et al., 2002). In Canadian rivers, Kostaschuk et al. reported surface suspended sediment concentrations reaching 1 g/L during the highest annual flows for the Fraser River delta in British Columbia (Kostaschuk et al., 1993). A maximum depth-averaged suspended particulate matter concentration of 3.5 g/L has been reported in the Bay of Fundy (Amos, 1996). A minimum three replicates of sediment-water mixture of 10 g/L sediment in 600 mL of 33‰ (parts per thousand) NaCl brine were prepared in 2.2-L fluorinated polyethylene bottles. In the case of kaolin and sand sediments, the pH of the sediment-salt-water mixture was adjusted to 7.0. Next, the mixtures were allowed to equilibrate for 20 min at 15° C. A 30-mL portion of oil was then added to the vessels (1:20 v/v oil: salt water). The mixing bottles were then sealed and allowed to thermally equilibrate for at least 4 hours at 15° C. The vessels were mixed overnight (16 hours) on the rotary end-over-end mixer unit (Associated Design & Manufacturing Co, Lorton, VA) at 55 rpm and 15°C. Then the contents were poured into a 1-L glass graduated container to settle for 24 hours at 15°C. The settling of the oil in the mixture was monitored during the 24 hour settling time and pictures were taken at specific time intervals to evaluate the settling velocity of the oil-sediment mixture. After 24 hours settling, where applicable, the mixtures were separated into three layers: a floating layer, a water layer and a settled layer. Subsamples were taken from each layer via different methods of pumping, scooping or pipetting, for further analysis. Care was taken to prevent disturbing the formed layers. The OPAs and floating layer were also evaluated in case of physiochemical properties analyses; including density, water content, micrography and particle size distribution. All of the tests were repeated for the blank control without the presence of sediment. For the blank (no sediment) tests all of the steps were the same as the original tests except for adding the sediment and the pH adjustment.

The density of oil-particulate aggregates (OPAs) were evaluated by examining the buoyancy of the OPAs in different concentration ratios of the mixtures of water: ethylene glycol or water: glycerol. For this experiment, initially a range concentrations of water: ethylene glycol (EG) (e.g., 100:0, 10:90, 90:10, 80:20, 70:30, 60:40, 50:50, 40:60, 30:70, 20:80, 10:90, 0:100 v/v) were prepared and their densities were analyzed as reference. 10 mL of each of the concentrations were placed in test tubes. Then 1 mL of OPA was added to the top of each test tube and the mixtures were left to reach equilibrium for 24 hours. Depending on the density of the OPA sample and water: EG mixtures, samples tend to have positive buoyancy in the mixtures with higher density, and negative buoyancy in mixture with lower density. In case that the sample is denser than high concentrated ethylene glycol, the water: glycerol mixtures are used instead. When a change of buoyancy of OPA is observed with the change of water: EG or water: glycerol mixture concentration, it can be concluded that the density of the mixture at the transition point is close to that of the OPA. Figure 1 illustrates an example of density evaluation for IFO 180-salt water-kaolin OPA with a sharp transition at water: EG 40:60 (v/v). The flask closest to the transition point (in the example of Figure 1, 40% water: 60% ethylene glycol) was chosen as the best match for density of the mixtures. Next, the content of that tube was filtered using 17 mm syringe filters with 0.45 μm pore size. Then, the density of filtrate liquid was measured in triplicate using Anton Parr DMA 5000 digital density meter (Anton Paar, Austria) for the final density evaluation.

Microscopic images of oil-particulate aggregates were observed and recorded using a Carl Zeiss AXIOSKOP (Carl Zeiss Canada Ltd.) equipped with a Zeiss Axiocam camera through a 40× magnification objective (Plan Fluor 40). After 24 h settling, a subsample was removed from the desired layer containing oil-particulate aggregates and diluted in 33‰ salt water at a ratio of 1:9 v/v sample: salt water to lower the concentration of suspended particles for better observation of the morphology. Observations were made in the transmitted-light mode illuminated with a halogen lamp as well as in epi-fluorescence mode using an X-Cite® 120 LED illumination system (Carl Zeiss Canada Ltd.). The fluorescence filter sets excitation= G 365 nm, beam splitter= FT 395 nm, long pass emission band pass= LP 420 nm.

The interaction between the oils and the sediments was followed by size distribution measurements using a Mastersizer 2000 particle size analyzer (Malven Instruments Ltd., Worcestershire, UK). The Mastersizer 2000 system includes an optical bench and a wet dispersion unit Hydro 2000 MU, which is used to prepare the sample and deliver it to the measuring cell in the optical bench. Samples were taken from the oil-particulate aggregates layer and added directly into a 1-L beaker which contains the dispersant (DI water). The distribution parameters, D₁₀, D₅₀ and D₉₀ are standard percentile readings, which respectively define the diameter below which 10%, 50%, and 90% of the overall particle/droplet volume is located. The D₁₀, D₅₀ and D₉₀ together with the volume mean diameter (VMD) are used in this study to describe the size of the particles in various oil-sediment dispersion systems.

The main focus of this study is characterizing the properties, behavior, and fate of fresh, moderate and heavily weathered CLB, as well as a comparison of oils, in interaction with sediment-salt water media. An Alberta sweet mixed blend (ASMB) was used as a light reference crude oil. An intermediate fuel oil (IFO 180) and a heavy fuel oil (Bunker C) were also included for comparison.

The evaluation of interaction of sediment-salt water with oil was performed in presence of 10 g/L sediment in 33‰ (parts per thousand) NaCl brine.

In case of kaolin and sand sediments, the pH of salt water-sediment mixture was around 5.7 and 6.5, respectively. As a result, before addition of the oil, the pH of salt water-sediment mixture was adjusted to 7.0 to be closer to the natural pH. In case on KA04 sediment, the pH of salt water-sediment mixture was around 8.5. Our evaluation showed that adjustment of pH to 7 did not have any significant effect in the results compared to the original pH. Therefore, the pH of the salt water-KA04 mixture was not adjusted for the rest of the study. The interaction of oil-salt water-sediment mixtures were simulated using a rotary end-over-end mixing unit for 16 hours at 55 rpm and 15°C. This condition has been shown to provide a high-energy mixing environment, with sufficient mixing time to ensure that if a mixed water-oil state is possible, it will likely be reached during the test (Fingas et al, 1998; Fingas and Fieldhouse, 2006). As mentioned earlier, after 16 hours mixing, the mixtures were poured into graduated containers and allowed to settle for 24 hours at 15°C and then the layers were separated for studying the physiochemical properties. The summary of our findings for oil-salt water-sediment and oil-salt water (no sediment) interactions is shown in Table 1 and Figure 2 and will be discussed in next section.

In case of the study without presence of sediment, the oil in the salt water mixture was emulsified and emulsified floating layer on top of the salt water column was observed. Four emulsion categories (stable, meso-stable, entrained water, and unstable) were defined by Fingas et al. based on water content and the visual appearance of the emulsion (Fingas et al, 1998; Fingas and Fieldhouse, 2006). The meso-stable state was observed for the lighter samples such as CLB fresh and ASMB oil, the unstable emulsion was formed for heavily weathered CLB (W4) and the entrained state was observed for the rest of oils. In general, positive buoyancy of oil was observed in absence of sediment, even for the heavily-weathered oil as the most viscous oil, indicating that evaporation alone is likely not sufficient to cause sinking of these oil in marine environments under the study conditions. Different oils interact differently with sediment-salt water mixture and as a result different behavior and buoyancy were observed, as can be seen in the Figure 2. In general, kaolin has a dispersive effect and the mixtures with kaolin formed finely dispersed oil-sediment particles in the water column. After the mixtures were poured into columns to settle, for most of the oils (CLB fresh, W2 and IFO-180), the oil-kaolin aggregates sank in the salt water with the exception of the highly weathered fraction (W4, weathered) and the light crude oil, ASMB (Figure 2A). The CLB W4 formed free-floating tarballs after 16 hours mixing with kaolin-salt water. The tarballs were stable only for 1–2 hours and then aggregated and converted to a floating entrained emulsion. In the case of ASMB, the dispersed OPA floated on top after 24 hours settling with no oil sinking to the bottom. This can be explained due to the much lower density of the OPA of the ASMB compared to density of salt water. When comparing the CLB fresh with CLB W2 (15.75%), the settled layer of oil-sediment is more compacted with the increase of % weathering. Compared to the kaolin, very similar results were observed for the KA04 sediment (Figure 2B), except for the CLB W4, which free-floating entrained emulsion was formed instead of tarballs (Figure 2C). In addition, the mixtures of KA04 with all fresh oils and the moderately weathered CLB (15.57%) were settled over time more quickly than kaolin. For both kaolin and KA04 oil-sediments interactions, apart from the sunken OPA layer, some small portions of oil remained floating on the salt water column. This floating layer is assumed to be the continuous free oil layer which contains some trapped sediment and water, and can be very easily disturbed or dispersed into the water column. For sand sediment interaction, very different results were observed compared to the kaolin and KA04 sediments. Except for the lighter oils (ASMB and fresh CLB), the rest of oils adhered on the walls of the vessel and were manually transferred to the cylinder for settling. The oils were emulsified and the majority of the oil remained floating on the surface of the salt water.

Due to the larger particle size of the sand (200–400 μm), it seems that there was no significant interaction between the oil and the sand. These results were very similar to the experiment in absence of sediment (See Table 4 and Figure 2D). As can be seen in Figure 2D, in case of interaction of sand with CLB W2, IFO 180 and Bunker C, small portion of the oils was found to be coated in a thin layer of sand and sunk to the bottom of the salt water column as larger blob-like particles. But they resurfaced after a while.

As already discussed, uptake of kaolin and KA04 resulted in OPAs for all oils except for CLB W4 (25.2%) which resulted in free-oil floating layer with trapped sediment and water. In contrast, the sand sediment did not result in the formation of an OPA layer for any of oil because of negligible interaction of oils with sand sediment and only a layer of free oil with trapped water and sediment was observed on top of the column.

Due to the aggregated or trapped sediment in the sample, the density analysis using a conventional density analyzer was not feasible. As a result, the density evaluation for these OPAs and free-oil floating layers for oil-sediment interaction studies were performed using in-house density analysis method which has been discussed in the “Materials and methods” section. In summary, the density analysis was performed via evaluation of the buoyancy of the OPAs in different concentration ratios of the mixtures of water: ethylene glycol or water: glycerol. Table 2 summarizes the key properties of the oils (including the density, particle size and buoyancy) after mixing with sediment-saltwater and saltwater only. Figure 3 shows the comparison of the density of the original oil, the OPAs or free-oil floating layers as a result of interaction with different sediments, and oil-in-water emulsion (no sediment). As can be seen in Table 2 and Figure 3, there is an increase in the density of oil with the formation of OPA. The oils with higher viscosity tended to have higher densities. The density evaluation for the OPAs of oil-KA04 interaction resulted in a range of densities rather than a single point, as indicated by vertical ranges on Figure 3. This was due to the gradual transition in the buoyancy of the sample in different concentration ratios of the mixtures of water: ethylene glycol or water: glycerol.

Figure 4A and 4D show the microscopic images of the two sediments kaolin and KA04 suspended in salt water under the transmitted-light mode. No fluorescence was observed for these two sediments under epi-fluorescence illumination. It is obvious that the kaolin consists mostly of very fine particulate while KA04 contains more coarse grains. Figure 4B and 4C illustrate the micrographs of CLB fresh-kaolin aggregates recorded at the same position in transmitted-light mode (Figure 4B) and fluorescence mode (Figure 4C). The use of both modes can provide complementary information since oil shows strong natural fluorescence, which derives largely from the aromatic fraction in the oil, while the sediment particles are only visible in the transmitted light unless they are coated with oil. Under fluorescence mode, oil droplets in spherical shape are clearly visible. Oil-sediment interaction is observed from the presence of oil droplets encrusted with sediment particles, shown as a spherical shape with rough surface. It is also noted that the oil droplets are interconnected by sediment particles to form droplet clusters, which is well illustrated in the micrographs under the transmitted light. These dispersions show good long-term stability as the droplets and droplet clusters remain dispersed over weeks and months. It is well known that solid particles can facilitate the stabilization of oil droplets in water when adsorbed to the oil-water interface.

The mechanism behind emulsion stabilization with particles is that the adsorbed particles form a dense film around the dispersed droplets, providing a mechanical barrier to rupture and coalescence. In our studies, high-energy mixing helps to break oil into droplets, suspend sediment particles, and enhances the collision between oil droplets and sediments. The solid particles act as stabilizer through adsorption at the droplet surface, sterically hindering coalescence of droplets. In addition, particle-particle interaction further contributed to droplet stability by forming three-dimensional network structures displayed as droplet clusters. Based on the microscopic examination, no significant differences in the morphology of the oil-kaolin aggregates are observed between different oils. Size differences are not obvious as well due to the coexistence of droplets of different sizes in each sample; however, the particle size measurement in the next section will be able to reveal the size variations for different oil-particulate aggregates. The microscopic images of oil-KA04 aggregates for CLB fresh are shown in Figure 4E (transmitted-light mode) and Figure 4F (fluorescence mode). Compared to the dispersions formed from oil-kaolin interaction, it is clear that the presence of KA04 sediment particles leaded to the formation of larger oil droplets. Moreover, although oil-sediment interaction is still observed, the oil is more likely to form distinct droplets with less clustering and network formation, indicating that the natural sediment KA04 interacts with the oils to a lesser extent as compared to kaolin particles. Unbound sediment particles were also present in the OPAs dispersions. These particles were settled at the bottom of the microscopic cell due to higher density and are not observed in shown images. In the cases of experiment with presence of sand, as well as experiment with no sediment present, no significant OPAs layers were observed and majority of the oil formed an oil-water mixture floating on top of water.

Particle size distribution was measured to reveal the size variations of the oil-particulate aggregates dispersions. It was known that these dispersions consist of a mixture of OPAs, oil droplets and free sediment particles dispersed in water. However, the Malvern Mastersizer system cannot discriminate between oil droplets, small sediment particles and oil-particle aggregates, and will measured all dispersed materials as particles.

For comparison, sediment-only dispersions were prepared by suspending kaolin, KA04 and sand in water, and the size distributions of the particles were measured. As illustrated in Figure 5, kaolin shows a multimodal distribution with D₁₀ of 0.2 μm, D₅₀ of 3.3 μm, D₉₀ of 11.73 μm, and VWD (volume mean diameter) of 5.8 μm. While size analysis of the natural sediment KA04 shows one predominant peak with D₁₀ = 3.1 μm, D₅₀ = 11.5 μm, D₉₀ = 42.9 μm and VMD = 18.9 μm. The natural benthic sediment contains particles mainly in the silt-sized range (90% in the range between 2–63 μm). Sand has much larger sized particles showing in one single peak ranging from 200 to 400 μm. Volume based particle size distribution (PSD) curves for the OPA layer from oil-kaolin interactions are shown in Figure 5A for CLB fresh and CLB W2, and in Figure 5B for ASMB, IFO-180 and Bunker C in comparison with the distribution of kaolin only. It was observed that the size distributions of oil-kaolin aggregates all showed one dominant peak, which is remarkably different from that of kaolin particles. The characteristic peaks of kaolin are not observed in oil-kaolin aggregates curves, indicating that most of the kaolin particles interact with the oil through adsorption at the oil-water interface around oil droplets. In addition, from the microscopic images, we observe that the majority of the individual oil droplets have diameter less than 20 μm, however, in the size distribution curve, a significant percentage of particles measured has diameters larger than 20 μm, especially for the weathered oil CLB W2 and heavy fuel oil Bunker C, which have over 50% of particles larger than 20 μm. The presence of large particles may be attributed to the formation of clusters of oil droplets interconnected by the particles. Figure 5C and 5D compared the size distribution of oil-KA04 aggregates. All the oil-KA04 aggregates dispersion show one simple peak shifting to larger particle size as compared to KA04 only, except for the moderately weathered oil CLB W2-KA04. This oil-sediment dispersion resulting from interaction between the CLB W2 and KA04 demonstrated a bimodal size distribution. The first peak at around 10 μm could be attributed to the presence of free KA04 particles, which may suggest that KA04 particles are not as strongly attached to the interface of the CLB W2 droplets as the kaolin particles. For the other oil-KA04 aggregates, no similar pattern was observed; however, it could be due to the closeness of particle size of the free sediment and the oil-droplets, resulting in overlap of the peaks.

Comparing the size variations for different oils, we can see that the peak value in the size distribution curve for the moderately weathered CLB (W2) shifted to larger size in comparison to the fresh CLB for both kaolin and KA04. Similarly, for the light crude oil ASMB, IFO, and Bunker C, the peaks position shifted to larger values following this order. Figure 6 compares the volume mean diameter (VMD) for oil-kaolin aggregates and oil-KA04 aggregates dispersions for CLB fresh, CLB W2, ASMB, IFO-180 and Bunker C. In general, oil with higher viscosity tended to form larger oil droplets and larger OPAs. Similar trends were observed for oil-kaolin and oil-KA04 aggregates for different oils, and the sizes of the oil-KA04 aggregates are consistently larger than those of the oil-kaolin aggregates for the same oil, which is in accordance with the previous findings by microscope observation. It can be seen that the average particle size of OPAs from CLB fresh is comparable to that from IFO-180, while the size of OPAs from CLB W2 is close to that of Bunker C. ASMB forms smaller OPAs when interacting with both kaolin and KA04, compared to the other oils.

In case of fine- and moderate-sized sediment, OPA dispersion was observed after high-energy mixing of abundant sediment particles with the CLB fresh, CLB moderately weathered (W2), IFO and Bunker oils that sank to the bottom of saltwater column. In case of ASMB (light conventional crude oil), the formed OPAs floated on top of water column after settling. The highly weathered oil CLB W4 did not uptake as much sediment, and instead formed discrete free-floating tarballs. When mixed with coarse sediment, all oils were emulsified and formed entrained or unstable-water-in-oil mixtures similar to those formed by oil-salt water mixing with no sediment present. However, large blobs of oil encrusted with sand were observed to sink in saltwater, these blobs were found to shed sand particles and would then resurface and coalesce into the floating surface oil. Microscopic examination showed that the formed OPAs consisted of oil droplets surrounded by sediment particles and were present in single droplet or multiple droplets clusters. The density and particle size analysis revealed that the OPAs of the oils with higher viscosity tended to have higher densities and larger particle sizes. When comparing kaolin and KA04, the sizes of the oil-KA04 aggregates were consistently larger than those of the oil-kaolin aggregates for the same oil, these findings were supported by microscope observation. In general terms, the OPAs formation and their buoyancy depend on the sediments' load and size as well as the type of the spilled oil and the state of weathering. Most likely, OPA formation would be more significant when the sediment sizes are fine enough (e.g., ≪200 μm as the case of this study) to interact with the oil under evaluation. Our research team is doing more evaluation on the formation of OPAs and their buoyancy base on a real oil spill scenario into fresh water containing sediments with different sizes.

This work was funded by the Government of Canada's World Class Tanker Safety Program. Authors would like to thank NRCan and Kinder-Morgan for providing oil samples. Authors are also thankful to ESTS Field Research Unit and DFO for sampling the natural sediments.