Search

Waste cutting emulsions have high oil concentration, complex components, and strong stability, making them extremely difficult to dispose of. Efficient separation of oil and water is key to disposing of such emulsions. In this study, magnetic nanoparticles (MNPs) were applied in different ways to the membrane separation process to form four coupling processes for emulsion treatment. Compared to direct ultrafiltration, the four coupling processes showed improved performance in chemical oxygen demand (COD) rejection and flux for sample waste cutting emulsions. In particular, the flux of the optimal process was 83 LMH, which was 41.5 times greater than the control process. The interactions between MNPs and emulsions, such as coalescence, coating, and migration under a magnetic field, were also studied. We believe that the introduction of MNPs actively affected the membranes; the MNPs avoided direct oil-membrane contact via the coating of oil drops and a cake layer that formed on the membrane surface, which then enhanced ultrafiltration and improved flux significantly.

O-linked glycosylation (O-glycosylation) and N-linked glycosylation (N-glycosylation) are the two most important forms of protein glycosylation, which is an important post-translational modification. The regulation of protein function involves numerous mechanisms, among which protein glycosylation is one of the most important. Core 1 synthase glycoprotein-N-acetylgalactosamine 3-β-galactosyltransferase 1 (C1GALT1) serves an important role in the regulation of O-glycosylation and is an essential enzyme for synthesizing the core 1 structure of mucin-type O-glycans. Furthermore, C1GALT1 serves a vital role in a number of biological functions, such as angiogenesis, platelet production and kidney development. Impaired C1GALT1 expression activity has been associated with different types of human diseases, including inflammatory or immune-mediated diseases, and cancer. O-glycosylation exists in normal tissues, as well as in tumor tissues. Previous studies have revealed that changes in the level of glycosyltransferase in different types of cancer may be used as potential therapeutic targets. Currently, numerous studies have reported the dual role of C1GALT1 in tumors (carcinogenesis and cancer suppression). The present review reports the role of C1GALT1 in normal development and human diseases. Since the mechanism and regulation of C1GALT1 and O-glycosylation remain elusive, further studies are required to elucidate their effects on development and disease.

With the rapid development of fluorescence microscope technologies, high-content screening and light-sheet microscopy are producing ever-larger datasets that pose great challenges in data storing and data sharing. As a popular compression tool, \(B^3D\) introduces a noise dependent compression algorithm for microscopy images to preserve the numerical intensities of all pixel within their uncertainties by exploiting the natural variability of each pixel value. Nevertheless, the high complexity of the processing flow restricts the deployment of the tool since the throughput and power consumption cannot satisfy the increasing demand. In this paper, we propose an efficient FPGA-accelerated data compression framework based on \(B^3D\). Following the co-design methodology, the compression processing flows are partitioned into different blocks to deploy on CPU or FPGA according to their computation characteristics. Also, we design a custom accelerator core that consists of multiple full on-chip pipelines using the channel function of the Intel OpenCL toolkit to implement data-flow driven computation. Our experiments show that the proposed framework achieves up to \(32\times \) throughput for a single pipeline compared with Intel Xeon E3-1220 v5 operating at 3.00 GHz, and \(6\times \) energy-efficiency compared with GPU implementation.

Demulsifying bacteria, as soft particles, are environmentally friendly agents used to recycle oil and purify water from surfactant-stabilized emulsions. However, details of this demulsification mechanism are unclear. Herein, we investigate the emulsion phase behavior and microstructural evolution in real space and time, as well as the specific role of the bacterium S-XJ-1 in this evolution through multifaceted approaches. Results reveal that new features of emulsion phase behavior with S-XJ-1 exhibited whole zone drop-drop coalescence, which caused emulsion network collapse, followed by drop rearrangement, which led to clear spatial layering of the emulsion. The bacteria adsorb onto the water-oil interface, aggregate into bacterial clusters, and become unevenly distributed at the interfaces in real emulsion systems. The bacterial peculiar behaviors were shown to improve drop-drop and drop-interface coalescence using micropipettes and sessile drop technology. Thus, the distribution and aggregation of bacteria at the interface could contribute to the bridging of neighboring water drops, rupture the original interface films, and improve drop coalescence. This study provides an insight into the role of particle demulsifiers in emulsion microstructural evolution.

Membrane separation has been acknowledged as an effective strategy to treat oily emulsions based on size-sieving. For separating nanoemulsions, smaller membrane pores are required to reject nano-sized oil droplets; however, the permeate flux would be significantly restrained. Herein, bioinspired by spider silk, a novel superhydrophilic electrospun nanofibrous membrane with spindle-knotted structures (SK-ENM) was constructed for highly efficient separation of nanoemulsions. The special spindle-knotted structures were constructed with 3–3.5 wt% polyacrylonitrile utilizing electrospinning based on Rayleigh instability. The prepared SK-ENM possessed a highly porous structure with a mean pore size of 0.634 μm, and strong superhydrophilicity/underwater superoleophobicity with rich micro/nano-hierarchical structures. For separating the surfactant-stabilized paraffin-in-water nanoemulsion with a mean droplet size of 325.7 nm, the oil rejection, permeate flux, and flux recovery rate were 95.68%, 1143.3 LMH/bar, and 99.98%, respectively. Thus, the SK-ENM with larger pores achieved both high oil rejection and high permeate flux for separating nano-sized oil droplets, which is a breakthrough achievement in overcoming the limitations of size-sieving. This could be attributed to the directional migration and coalescence of nano-sized oil droplets on spindle-knotted structures driven by Laplace force, which was calculated to be 8.7 × 10-8N. Meanwhile, the resulting large oil droplets were proved to be rejected by the superhydrophilic nanofibrous membrane with negligible fouling. This study provides insights into the design of novel functional superhydrophilic membranes for separation of emulsions. The as-prepared SK-ENM can be effectively applied to treating various nanoemulsions and complex oily wastewater in practical.

University of Technology Sydney. Faculty of Engineering and Information Technology. As a fundamental problem in query processing, similarity search has been applied in many fields including multimedia, machine learning, database, recommendation systems and so on. Generally, it will be challengeable when it comes to the high-dimensional space due to “the curse of dimensionality”. Since it would be too expensive to find exact results, approximate solutions have been studied in many papers. There are various distance metrics to evaluate the similarity between the query object and other points in a dataset. In this thesis, we focus on some well-known distance metrics including Euclidean distance and inner product. Except for Euclidean space, we also study query processing on graphs and propose a novel distance metric on graph. The thesis contains four similarity search problems regarding to different distance metrics, which are Approximate Nearest Neighbour, Approximate Inner Product Search, Approximate Furthest Neighbour and Skyline Nearest Neighbour. Given a query point, the four problems all focus on retrieving a set of “similar” points from the dataset. Given a set of 𝘥-dimensional data points, and a query point 𝘲, Approximate Nearest Neighbour Search (ANNS) aims to find the approximate closest object to 𝘲 in the set. More specifically, in this thesis, we focus on 𝘤-ANNS problem, which means given a constant c, the purpose is to find a result whose distance is not larger than 𝘤 times of the exact smallest distance with a certain possibility. Even though this problem has been researched for a long time, there are still some shortage of current algorithms. We studied the existed works, and proposed a novel I/O efficient algorithm to solve 𝘤-approximate nearest neighbour problem in external memory, which can dramatically reduce I/O cost and provide rigorous proof of its correctness. Maximum Inner Product Search (MIPS) is another valuable problem. It returns an object with maximum inner product value to query point 𝘲. There are hundreds of solutions for MIPS but still short of comprehensive evaluation and analysis of these methods’ performance. In this thesis, we chose several state-of-the-art algorithms of MIPS using different techniques, and conducted a set of comprehensive experiments to evaluate their performance fairly. Approximate Furthest Neighbour Search is an opposite problem of Nearest Neighbour Search. It finds the furthest object to query point 𝘲 in a dataset instead of the closest one. Since most recent works for approximate furthest neighbour search in external memory are only suitable for low-dimensional data, we proposed a new I/O efficient technique to achieve a better performance on I/O cost. In addition to Euclidian space, similarity search is also a fundamental problem in other spaces like graphs. Considering real-world applications, the multi-layer graph model is extensively studied to reveal the multi-dimensional relations between the graph entities. In this thesis, we formulated a new problem called skyline nearest neighbor search on multi-layer graphs, and proposed a baseline algorithm, and two optimizations instead of naively adopting the traditional skyline procedure as a subroutine. We also investigated the rule to optimize search order in the algorithm so that further improve the algorithmic efficiency.

Oil adsorption from oily wastewater using three-dimensional porous materials is a popular mode of separation, but the adsorption of oil droplets from oil-in-water (O/W) emulsions is difficult. In this study, hydrophobic melamine sponges, prepared by coating with dopamine (DA) and polyethyleneimine (PEI) via controlled copolymerization, were utilized to separate O/W emulsions on a self-designed “oil adsorption-water filtration” device. Unexpectedly, the sponge obtained via DA/PEI modification was highly hydrophobic. Molecular dynamics simulations revealed that this was due to Schiff base production upon copolymerization of polydopamine (PDA) and PEI on the melamine sponge. The sponge showed excellent adsorption capacities, 67.2–178.6 times its own weight, for various oils. More importantly, a surfactant-free O/W emulsion was separated with 93.5% separation efficiency and 24.9 g/g sponge oil adsorption. The sponge also exhibited perfect separation efficiency for an anionic-surfactant-stabilized O/W emulsion. During filtration, negatively charged oil droplets were captured on the amino-rich sponge by electrostatic attraction. Furthermore, the hydrophobic effect was found to be the driving force in adsorbing and collecting oil on the sponge skeleton while repelling the continuous water phase. Since the as-prepared sponge can simultaneously adsorb oil droplets and filter the water phase in O/W emulsions, it has major potential for application in oily wastewater treatment.