• Membrane Journal
• /
• v.33 no.1
• /
• pp.34-45
• /
• 2023

Two computational intelligence techniques namely artificial neural networks (ANN) and support vector machine (SVM) are employed to model the permeate flux based on seven input variables including time, transmembrane pressure, rotating velocity, the pore diameter of the membrane, dynamic viscosity, concentration and density of the feed fluid. The best-fit model was selected through the trial-error method and the two statistical parameters including the coefficient of determination (R²) and the average absolute relative deviation (AARD) between the experimental and predicted data. The obtained results reveal that the optimized ANN model can predict the permeate flux with R² = 0.999 and AARD% = 2.245 versus the SVM model with R² = 0.996 and AARD% = 4.09. Thus, the ANN model is found to predict the permeate flux with high accuracy in comparison to the SVM approach.

• Proceedings of the Membrane Society of Korea Conference
• /
• 1998.04a
• /
• pp.105-109
• /
• 1998

1. Introduction : In microfiltration the transport, deposition and removal of particles control cake formation on a filter. In this connection a new model on cake formation, based on the wall shear stress, was tested here in comparison with experiments of fine particle slurry under Taylor-vortex flow. The model expresses the deposition process for particles as two first-order steps in series of mass transfer and adhesion, and their removal process as a linear relation to the wall shear stress. This embraces characteristics of both dead-end and crossflow filtration. The correlation resulting from fitting to experimental data represented the experimental data reasonably well. This study will be helpful in analyzing fouling in heat exchangers.

• Membrane Journal
• /
• v.13 no.2
• /
• pp.88-100
• /
• 2003

A change of filtrate flux in Taylor vortex flow filtration was investigated experimentally by rotating speed of inner cellulose ester membrane cylinder (average pore size: 1.2 ${\mu}m$), slurry concentration, and particle size. The filtrate flux was a direct proportion relation with TMP, but an inverse relation with resistances. A change of cake resistance with time was examined by rotating speed, slurry concentration, and particle size. Initial resistance increased dramatically as raising slurry concentration, and the pseudo-steady state was maintained at high resistance value. However, times to reach the pseudo-steady state did not depend on slurry concentration. The resistance was larger as smaller particle size, because possibility of pore blocking inside membrane could be higher and shear effect should be lower as smaller particle size. A model equation suggested in this study was composed of particle deposition and removal terms, and could confirm well experimental data using average values of experimental coefficients.

• Journal of Korean Society of Environmental Engineers
• /
• v.22 no.7
• /
• pp.1173-1182
• /
• 2000

The rapid mixing process has been considered as an important step in water treatment. Since the coagulant dispersion into raw water by rapid mixer can influence on the flocculation and filtration efficiency, many researchers have developed various devices and mixing methodologies. Until now, they focused attention on only coagulant dose, pH. rotating velocity and G value but overlooked the real turbulent flow and mixer geometry in rapid mixer. Therefore this paper questions the significance of turbulent flows in rapid mixer and focuses on the analysis of turbulent fluid in various mixer geometry with CFD(Computational Fluid Dynamics). The results of the jar-tests using various geometries indicate that the turbidity removal rate in a circular jar without baffle is higher than that of a circular with baffle. And the turbidity removal rate in Hudson jar is also founded to be higher than in the circular jar with baffle. The CFD simulation of velocity fields in jar demonstrates that the differences of removal rates among the various geometries are largely due to the formation of the different turbulent fluids fields with different geometries.

• The KSFM Journal of Fluid Machinery
• /
• v.19 no.4
• /
• pp.48-53
• /
• 2016

The centrifugal separator which uses gravity separation method for oil-water separation, rotating at high-speed, is one of the most commonly used device for controlling the amount of the oil in waste water collected in bilge. The IMO (International Maritime Organization) has set regulations, also known as MARPOL 73/78, for the prevention of marine pollution. In addition, DET NORSKE VERITAS (DNV) has set standards regarding the assignment of Environmental Class Notation, CLEAN or CLEAN DESIGN, of ships. One of the requirements for classification is that in addition to conforming to MARPOL 73/78, more stringent measures must be taken as well. One of these measures is to limit the oil concentration in bilge water to less than 5ppm. So in this study, an Oil-Water Separator (OWS) is used together with multiple separating plates as a filtration system to be used as an oil-water separation device. The OWS operates using centrifugal separation in which the mixture is separated by centrifugal forces. The main purpose of this paper is to present the OWS separation efficiency according to the rotation speed, mass-flow rate, the angle and the number of stacked layers of the laminated plate using Computational Fluid Dynamics (CFD). Improvements to the device will be investigated from these results.

• Macromolecular Research
• /
• v.16 no.3
• /
• pp.212-217
• /
• 2008

PVDF ($Kynar^{(R)}$ 761) nanofibers were made by electrospinning with an external voltage of 6-10 kV, a traveling distance of 7-15 cm and a flow rate of 0.4-1 mL/h. Although the mean diameter of the fibers has not changed significantly, the conditions affected the change in diameter distribution. This was attributed to interactions, both attraction and repulsion, between the positive charges on the polymer solutions and the electrically grounded collector. Higher voltages and traveling distance increased the level of attraction between the positive charge on the polymer solution and the electrically grounded collector, resulting in a narrow diameter distribution, In addition, a high flow rate allowed a high population of uniformly charged solutions to travel to the grounded collector, which resulted in a narrow diameter distribution. The optimum conditions for electrospinning of PVDF in DMAc/acetone (3/7 by wt) were a collector voltage of 6 kV, a syringe tip to collector of 7 cm, a flux rate of 0.4 mL/h and 10 kV, 10 cm, 1 mL/h, Since PVDF is widely used as a filtration membrane, it was electrospun on a PET support with a rotating drum as a grounded collector. Surprisingly, some straight nanofibers were separated from the randomly deposited nanofibers. The straight nanofiber area was transparent, while the randomly deposited nanofiber area was opaque. Both straight nanofibers and aligned nanotibers could be obtained by manipulating the PET drum collector. These phenomena were not observed when the support was changed to an Al sheet. This suggests that a pseudo dual collector was generated on the PET sheet. No negative charge was created because the PET sheet was not a conductive material. However, less charge was created when the sheet was not perfectly attached to the metal drum. Hence, the nanotibers jumped from one grounded site to the nearest one, yielding a straight nanofiber.