2013 Progress Report: Affinity-Based Hydrocyclone Filter for Oil-Water Separation and Oil Spill CleanupEPA Grant Number: R835183
Title: Affinity-Based Hydrocyclone Filter for Oil-Water Separation and Oil Spill Cleanup
Investigators: Tarabara, Volodymyr , Bruening, Merlin , Bénard, André
Institution: Michigan State University
EPA Project Officer: Lasat, Mitch
Project Period: May 16, 2012 through March 31, 2015 (Extended to March 31, 2016)
Project Period Covered by this Report: April 1, 2013 through March 31,2014
Project Amount: $500,000
RFA: Environmental Impact and Mitigation of Oil Spills (2011) RFA Text | Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Ecosystems
The goal of this research project, as set in the proposal, is to develop a hybrid hydrocyclone-membrane technology that separates oil-water mixtures into a water stream that meets standards for discharge into the environment and an oil stream sufficiently dewatered for energy use. The four specific research objectives of the project are to: (1) create hydrophilic, regenerable coatings to mitigate membrane fouling; (2) develop tools for predictive and diagnostic modeling of hydrocyclone flow for performance optimization; (3) design superoleophilic membranes to continuously whicker oil; and (4) design and build a pilot scale crossflow filtration hydrocyclone for concept validation.
An outreach component of the project aims at engaging and connecting communities in Louisiana and Michigan that were affected by recent oil spills (BPs Gulf of Mexico spill and Enbridges Kalamazoo River spill). The objective of the outreach effort is to demonstrate that communities have a stake in research projects aimed at addressing the significant human health and environmental risks posed by oil spills.
Task 1: Design of hydrophilic, regenerable membrane coatings for efficient oil-water separation
We characterized oil-fouling reduction on polyelectrolyte-coated porous membranes used for oil-water filtration. Although polyelectrolyte multilayers with negatively charged surfaces appear to reduce the fouling caused by negatively charged oil droplets, results were not highly reproducible, and irreversible fouling occurred within minutes. Thus, we developed superhydrophilic films with water contact angles of below 10° by layering positively charged polyelectrolytes with negatively charged silica nanoparticles. By varying the number of layers, as well as the size of the silica nanoparticles, we created superhydrophilic films with varying thicknesses. The films were oleophobic, with oil contact angles above 100° in water, and rejected nearly 100% of the oil in filtration of oil water-mixtures. Unfortunately, as with the simple polyelectrolyte coatings, severe fouling occurred within minutes of beginning filtration.
Polyelectrolyte brushes (Figure 1) show more promise than multilayer polylectrolyte films for reducing membrane fouling. Atom-transfer radical polymerization enabled controlled growth of poly(sulfopropyl methacrylate-potassium salt) (SPMK) from immobilized initiators to allow variation of the brush thickness by changing the polymerization time. In dead-end filtration of oil-water emulsions through nylon membranes modified with SPMK brushes, surfactant charge greatly affects flux and rejection.
Figure 2 shows that membranes coated with the polyanionic brush completely reject oil droplets stabilized by sodium dodecyl sulfate (SDS, an anionic surfactant). Moreover, the permeate flux, which is only slightly lower than that of pure water, is constant with time, suggesting that the electrostatic repulsion of negatively charged droplets prevents membrane fouling. In contrast, in the presence of cetyltrimethylammonium bromide (CTAB, a cationic surfactant), oil droplets permeate through modified membranes easily, and the membrane permeability initially increases with time. Most likely, the anionic polymer brushes absorb cationic surfactants and become less hydrophilic. This may lead to brush collapse (deswelling) and a larger effective membrane pore size. To verify this hypothesis, we are currently studying interactions between surfactants and polymer brushes.
Additionally, we investigated how the thickness of polymer brushes affects fouling and flux through the membrane. We found that membranes modified with brushes polymerized for 15 min have high oil rejection along with higher permeate flux than membranes with brushes polymerized for 1 h. However, longer polymerization times can make the brush more hydrophilic and less susceptible to fouling. Thus, we need to find a balance between high permeate flux and fouling resistance.
Task 2: Computational modeling of liquid-liquid separation crossflow filtration hydrocyclone (LL-CFFH)
Hydrocyclones and membranes are among the most important technologies currently used for produced water treatment. In this project, we developed three types of swirling flow devices with improved performance: (1) hydrocyclones with chambers of redesigned shape, (2) crossflow filtration hydrocyclones hybrid devices that combines both hydrocyclones and membranes, and (3) rotating tubular membranes. We have studied numerous configurations using computational fluid dynamics (CFD). Some of the CFD predictions were verified experimentally. Various CFD simulations are discussed below.
2.1: CFD Study for the redesign of hydrocyclones for improved performance
In this aspect of the project, we examined the internal flow structures within hydrocyclones used for liquid-liquid separation, especially those used for the removal of oil droplets from water. The internal flow structures and patterns are greatly influenced by the geometric shape of the swirl chamber. The effects of parabolic and hyperbolic wall profiles of the swirl chamber (Figure 3) on the reverse flow vortex core, short circuit flows, and the separation efficiency are investigated numerically by solving the Reynolds Average Navier-Stokes equations closed by an equation of change for the Reynolds stress. Results of this research will be presented in Motin et al., 2014 [IMECE2014-37190]. Droplet trajectories are predicted by solving a kinematic equation of motion and force balance. Internal flow structures for different geometric conditions have partially motivated the redesign of the hydrocyclone geometry so as to support a longer and stable reverse flow vortex core and for greater separation efficiency as shown in Figure 4. Results indicate that both the parabolic and hyperbolic swirl chambers provide improved separation efficiency. However, the hyperbolic swirl chamber has a greater potential for the reduction of effective length of the hydrocyclone with maintaining the same separation efficiency (Figure 4).
2.2: CFD study of hydrodynamics and separation performance of a crossflow filtration hydrocyclone (CFFH)
Various hydrodynamic aspects and the separation performance of a novel CFFH have been studied using CFD. A CFFH is a device that combines the desirable attributes of a crossflow filter and a vortex separator into one unit to separate oil from water. Such a device is shown in Figure 5. The velocity and pressure fields within the CFFH are estimated by numerically solving the filtered Navier-Stokes equations (by using a Large Eddy Simulation (LES) approach). The Lagrangian approach is employed for investigating the trajectories of dispersed droplets based on a stochastic tracking method called the Discrete Phase Model (DPM). The mixture theory with the Algebraic Slip Model (ASM) is also used to compute the dispersed phase fluid mechanics and for comparing with results obtained from the DPM. In addition, a comparison between the statistically steady state results obtained by the LES with the Wall Adaptive Local Eddy-Viscosity (WALE) subgrid scale model and the Reynolds Average Navier-Stokes (RANS) closed with the Reynolds Stress Model (RSM) is performed for evaluating their capabilities with regards to the flow field within the CFFH and the impact of the filter medium.
Effects of the Reynolds number, the permeability of the porous filter, and droplet size on the internal hydrodynamics and separation performance of the CFFH are investigated. Results indicate that for low feed concentration of the dispersed phase, separation efficiency obtained based on multiphase and discrete phase simulations is almost the same. Higher Reynolds number flow simulations exhibit an unstable core and thereby numerous recirculation zones in the flow field are observed.
Results have been summarized in Motin et al., 2014 (Journal of Membrane Science; in review) and show that improved separation efficiency is observed at a lower Reynolds number and for a lower permeability of the porous filter.
2.3: CFD study of the separation performance and hydrodynamics of a rotating tubular membrane filtration system
The performance of a liquid-liquid separation process based on an axially rotating tubular ceramic membrane operated in a crossflow regime as shown in Figure 6 is studied numerically with oil-water dispersions used as a model mixture. Internal hydrodynamics are explored using CFD simulations to obtain the velocity field in the continuous phase (water) and predict the separation efficiency with respect to the dispersed phase (oil). A discrete phase model is used to estimate trajectories of dispersed oil droplets within the membrane channel. The separation performance of the process is evaluated in terms of the droplet cutoff size. Effects of the Reynolds and Swirl numbers on velocity and pressure fields, shear stress, droplet cutoff size, and separation efficiency are investigated. The increased shear stress on the membrane surface due to the angular and the crossflow velocities decreased the accumulation of droplets on the membrane while increasing the separation efficiency. The droplet cutoff size is observed to decrease with an increase in the Reynolds and Swirl numbers. The separation efficiency strongly depends on the Swirl and Stokes numbers but only weekly on the Reynolds number. By increasing the Swirl number of the flow, it may be possible to remove very fine droplets by centrifugal force only and avoid membrane fouling.
Task 3: Design of superoleophilic membranes to continuously whicker oil
We previously prepared superhydrophobic coatings to wicker oil, but thus far they were not effective under the pressures required for filtration. Thus, we abandoned this aim.
Task 4: Design of the prototype hydrocyclone system
We have assembled hydrocyclones of three different types: (1) a solid-liquid separation hydrocyclone (SL-H) for separating dispersed phases heavier than water; (2) a liquid-liquid separation hydrocyclone (LL-H) for separating oil and water; and (3) a liquid-liquid separation crossflow filtration hydrocyclone (LL-CFFH) for separating oil and water where a significant fraction of oil exists as small (< 30 μm) droplets that cannot be efficiently removed by a LL-H. In Task 4.1, we experiment with complex three phase (oil, water, sand) systems while Task 4.2 focuses on liquid-liquid crossflow filtration hydrocyclones.
4.1: Experimental design and testing of SL-H
Simple two-phase feeds such as water-sand or water- oil are normally not representative of the actual multiphase nature of produced water. To better model produced water, in a subset of our experiments we used feeds that contained water oil and sand. A 22 mm solid-liquid hydrocyclone was manufactured for our team by ChemIndustrial Systems, Inc. (CSI), Cedarburg, WI www.chemindustrial.com) in 2013. Experimental investigation of oil-sand-water interactions using the SL-H began in July 2013 and are ongoing. Our aim is to investigate hydrocyclonic separation of a water-oil-sand mixture and evaluate the impact of agglomeration on the separation efficiency.
In a hydrocyclone separation system with no interaction between the components, the lighter phase (oil) tends to leave the system with the overflow stream while the heavier phase (sand) settles tends to leave through the underflow. In systems where phases interact, the above behavior can be expected to be altered. These interactions in the system can form agglomerates of three types: (a) sand particles with oil adsorbed on their surface; (b) oil droplets with smaller sand particles attached on their surface; and (c) oil-sand agglomerates with comparable volumetric fractions of sand and oil. Agglomerates of the first type will likely have an overall density greater than water and preferentially be removed through the underflow; the second type of agglomerates will likely have a density close to that of oil and be preferentially removed through the overflow. The feed water used in our tests included 1,000 mg/L of suspended sand with a specific gravity of 2.6 and 1,000 mg/L of emulsified mineral oil with a specific gravity 0.86. Figure 7 shows the confocal microscopy image of agglomerates in the mixture.
To evaluate the effectiveness of the separation of water-oil- sand mixture we compared: efficiencies of separation of oil from the water-oil system and from the water-oil-sand systems, as well as efficiencies of separation of sand from these two types of feed. Separation efficiencies are defined as follows:
As shown in Figure 8, the separation efficiency of oil does not show a statistically significant dependence on the presence of sand in the feed. The separation efficiency of sand, however, decreases when oil is present. We interpret this change as due to the removal of oil-associated small particle of sand through the overflow.
4.2: Experimental design and testing of liquid-liquid separation hydrocyclone (LL-H) and liquid- liquid crossflow filtration hydrocyclone (LL-CFFH)
The LL-H and LL-CFFH (Figure 9) were also manufactured for the project by CSI at the overall cost of $2,210 and were received by the MSU team on June 28, 2013. After a month of assembly and configuring of the data acquisition system, the oil-water experiments began in August 2013. As should be clear from Figure 9, the LL-H (a) can be converted to a LL-CFFH (b) by replacing the tailpipe with a membrane, allowing direct performance comparison.
Initial experiments were performed using the LL-H, with soybean oil (919 g/L) and mineral oil (860 g/L) used to prepared model oil-water emulsions. Concentrated emulsions were prepared and diluted to 1,000 mg/L. Testing was performed at various flow rates, pressures, and split ratios to determine the optimum operating parameters based on the oil separation efficiency. Figure 10 presents efficiency data for mineral oil as a function of the split ratio for membranes of three pore sizes.
The separation efficiency of the LL-CFFH was measured using the same procedure as in tests with the LL-H. The permeate stream from the membrane unit was recycled back to the feed tank to maintain the feed concentration of oil at an approximately constant level. The separation efficiency of the two processes (LL-H and LL-CFFH) as a function of the volume of treated water is shown in Figure 10. This experiment was repeated for multiple membrane pore sizes. The separation efficiency was recorded to be in the 37% to 52% range. The results showed the substitution of the hydrocyclones tailpipe with a membrane did not detrimentally affect the hydrocyclones performance. The fact that the CFFH with the largest pore size membrane showed the highest ��oil was an unexpected result that gives hope that practically important larger permeate flow rates can be achieved at the same time with acceptable quality of treated water.
Membrane fouling tests were performed with an oil-water emulsion to determine the effect of the hydrocyclone on the membrane performance. It was hypothesized that membrane fouling in the LL-CFFH process would be reduced due to two separate mechanisms. First, because the hydrocyclone directs a fraction of oil (equal to 1 − ��oil) into the overflow stream, the membrane in the tailpipe region is exposed to a lower amount of oil. Second, centripetal forces acting on oil droplets that do enter the membrane channel will move this oil towards the centerline of the stream and away from the membrane surface.
Three set of experiments were used to determine the effect of each mechanism: (1) Crossflow filtration with Coil = 1,000 mg/L; (2) Crossflow filtration with Coil = 600 mg/L; and (3) LL-CFFH with Coil = 1,000 mg/L. The same membrane was used in the crossflow and LL-CFFH tests. The difference in fouling between experiments 1 and 3 can be attributed to both mechanisms acting simultaneously. Because the hydrocyclone removes approximately 40% of the oil (Figure 10), the difference in fouling between experiments 2 and 3 should be due to fluid rotation in the membrane channel caused by the hydrocyclone, as both membranes in the crossflow filtration test and in LL-CCFFH tests - are exposed to the same concentration of oil. The results showed the membrane was fouled least during LL-CFFH experiments (Figure 11) and that the reduction of fouling is a synergistic benefit of combining the membrane and the hydrocyclone in a single LL-CFFH unit.
A presentation was made at the 2014 No-Spills Conference (24th) in Traverse City, Michigan concerning the project and outreach plan. Andre Benard provided a project overview and update to the Louisiana Sea Grant Marine Extension Program faculty at their quarterly meeting in Baton Rouge, Louisiana.
Polymer brushes are attractive for decreasing fouling, but their performance depends on surfactant charge and brush thickness. In the coming year, we will employ in situ ellipsometry to examine when surfactant-brush interactions cause brush collapse. We will also carefully control the polymerization time on polymeric membranes to maximize flux while employing brushes to minimize fouling. Finally, while employing charged surfactants, we aim to create a separation membrane module with electrodes to apply an electric field during filtration. The field should promote demulsification of an oil-water mixture, and through electrostatic effects force charged oil droplets away from a membrane surface.
Simulations will continue with the goal to:
- Identify geometric parameters that maximize the performance of LL-H and LL-CFFH. Parameters to vary will include the wall taper angle for a conical hydrocyclone, the various lengths of each sections, the inlet velocity (or Reynolds number), the pressure drop, the applied back pressure on the underflow;
- Validate the computational results with experimental data for the performance of the LL- CFFH;
- Identify a geometry that will maintain a strong swirl within the membrane channel. This may require several inlets and outlets as the swirl dies rapidly in the presence of a porous wall. The swirl is needed to mitigate fouling of the membrane.
This task has not proven fruitful and is no longer a focus of the work.
We will complete the assembly of the LL HC and will test it with model oil-water emulsion and samples of produced water. Further, we plan to work with ChemIndustrial to explore the potential of miniclones (http://www.hydroclone.com/products/miniclone.htm). These devices seem to be very promising because of the possibility of arraying a large number of them and because these devices afford the highest centrifugal acceleration of all hydrocyclone models. The arraying may help to overcome perhaps the most serious drawback of CFFHs low permeate flow rate from the microfilter. High acceleration in the miniclone geometry should allow separation of very smaller oil droplets. We consider submitting an SBIR proposal (with CSI as the lead and the MSU team as a sub-constructor) on miniclones for oil-water separation.
The website is under development and will be made available to the public in the final year of the project. Website sections (sub-pages) proposed include: Home, People, Publications, Resources, News (pictures, research updates), and Contact Us. As a part of Resources, we will include Background on Enbridge & Deepwater Horizon spills, which will have a table with the following possible entries: Amount of oil spilled, Date of the spill, Receiving water body, Oil type, etc. An Our Technology page may also be produced once we are ready to share details concerning the technologies utilized and developed in the project with the general public.
A comprehensive PowerPoint presentation is being developed outlining the project, progress made thus far, and targeted outcomes. Louisiana Sea Grant will assist in producing a companion video to go along with the website and PowerPoint. Community organizations in oil-spill affected Louisiana and Michigan areas will be identified and invitations extended to the leaders to have the outreach team deliver presentations to the organizations. Survey instruments will be developed and used to obtain feedback from the community organizations contacted.
Prospective businesses that may be interested in licensing the oil-water separation technology will be identified and contacted concerning possible adoption and adaptation of these technologies in current and future oil spill cleanup operations.
The Gulf of Mexico Research Initiative (GoMRI) has been funded and charged with creating an independent research program designed to study the impact of the BP oil spill on the environment and public health in the Gulf of Mexico. They will be made aware of our project and technology we are developing. We consider GoMRI as a possible source of funding to continue our work after EPA STAR funding ends.
Journal Articles on this Report : 2 Displayed | Download in RIS Format
|Other project views:||All 43 publications||5 publications in selected types||All 5 journal articles|
||Ji P, Motin A, Shan W, Benard A, Bruening ML, Tarabara VV. Dynamic crossflow filtration with a rotating tubular membrane: using centripetal force to decrease fouling by buoyant particles. Chemical Engineering Research and Design 2016;106:101-114.||
||Motin A, Tarabara VV, Benard A. Numerical investigation of the performance and hydrodynamics of a rotating tubular membrane used for liquid–liquid separation. Journal of Membrane Science 2015;473:245-255.||