Dead-end Ultrafiltration Research Articles

A comparative study between classical stirred and ultrasonically-assisted dead-end ultrafiltration

The aim of this work was to determine mass transfer coefficients in the cases of ultrasonically-assisted and classical stirred dead-end ultrafiltration. A comparative study was then performed, and mass transfer coefficients obtained under ultrasonic conditions are described by an empirical model. This correlation results from an analogy with what is observed using a stirred cell and involves the ultrasonic power as the main parameter. The hydrodynamics are assumed to depend on the intensity of the ultrasound effects illustrated by the agitation arising within the cell. This agitation is due to convective currents as well as physical effects due to cavitation. The concentration polarization phenomenon is therefore affected by this action of ultrasonic waves.

Low-Frequency Ultrasound to Improve Dead-End Ultrafiltration Performance

Low-frequency ultrasound (20 kHz) has been used to facilitate dead-end ultrafiltration of polymer solutions. The influence of ultrasonic irradiation on permeate fluxes and retention ratios was investigated and compared with conventional stirred and unstirred operations. In addition, the effect of the ultrasonic field on the membrane permeability and on the polymer solution was examined. The efficiency of ultrasound in enhancing ultrafiltration performance was demonstrated. This improvement is due to a decrease of the boundary layer resistance against permeation flow, which may be attributed to disturbance induced within the cell and at the membrane surface by physical effects associated with ultrasonic waves propagation.

Feed Cycling in Dead End Filtration

The dead end ultrafiltration of dispersions of active carbon in water was studied. High flux of the permeate and effective purification were goals of the studies. Cyclical flow of the feed (cycling) was found to be the best condition for good purification. During the purification of the membrane, an air/water dispersion was injected into the hollow fiber membrane (air sparging) to improve the cleaning effect. A Simplex optimization of significant parameters was carried out. The behavior of pressure drop and module entrance pressure permitted the classification of the values into two groups. A different build-up of the cake of these groups during the filtration is possible.

Feed Cycling in Dead End Filtration

The dead end ultrafiltration of dispersions of active carbon in water was studied. High flux of the permeate and effective purification were goals of the studies. Cyclical flow of the feed (cycling) was found to be the best condition for good purification. During the purification of the membrane, an air/water dispersion was injected into the hollow fiber membrane (air sparging) to improve the cleaning effect. A Simplex optimization of significant parameters was carried out. The behavior of pressure drop and module entrance pressure permitted the classification of the values into two groups. A different build-up of the cake of these groups during the filtration is possible.

On the use of dead-end ultrafiltration for the study of the influence of sulfate and/or silicate on particle size distribution in concentrated and partially prehydrolyzed Al(III) flocculants

Because of their great tendency towards hydrolysis, concentrated and partially prehydrolyzed Al(III) flocculants cannot be diluted as required by many physico-chemical characterization methods. In this work, the complete dead-end ultrafiltration has proven to be a useful characterization method for these products. The results presented have shown that the behavior of the membrane/solution interphase could be described by general ultrafiltration (UF) theories. In the first part of this work, the most discriminating UF method (serial UF) is chosen for the characterization of concentrated flocculants. Next, prehydrolyzed flocculants with an increasing silicate and/or sulfate content were synthesized and the influence of silicate and/or sulfate on particle size distribution was resolved by serial UF. The silicate content is correlated to the size increase of the colloidal particles in such flocculants. This is explained by the reduced effectiveness of fragmentation of aluminosilicate particles formed during flocculant synthesis as the silicate content increases. Sulfate also increases the mean size of the particles, but the mean size is not correlated to the sulfate content. The role of sulfate is apparently to prevent the dissolution of the Al hydroxide particles formed in the initial step of flocculant synthesis. UF in presence of silicate and sulfate shows that, in such cases, sulfate could also limit the aluminosilicate fragmentation leading to a greater increase of the particle size than that observed with only silicate addition.

Dead-end ultrafiltration of pretreated and untreated WWTP effluent for re-use in process water applications

Ever rising costs of water supply due to tax-rules, more stringent quality requirements and the concern of the continuing depletion of ground water sources give strong impetus for new innovative technologies. These technologies must be reliable and economically attractive. Recently, the membrane manufacturer X-Flow, a full subsidiary of the Norit Group, has developed a novel 8-inch dead-end ultrafiltration technology. This technology has proven to be a viable alternative for the economical production of water in a number of applications e.g. re-use of back-wash water of mediafiltration and filtration of surface water (lakes, rivers, canals). A number of full size plants has been realized or is under construction at this moment. Norit Membrane Technology has started experiments in 1997 in which dead-end ultrafiltration, using the 8-inch technology, was used in the reuse of Waste Water Treatment Plant (WWTP) effluent. Experiments have been carried out with and without pretreatment. Pilot scale installations (1–6 m 3h −1) were used for the experiments. Preliminary results show a very good permeate quality with an MFI of 0.5 to 0.6, making this technology especially suitable as reverse osmosis pre-treatment. Permeate fluxes of up to 80 lm −2h −1 for longer periods of time were measured and cleaning of the membranes posed no significant problems. Pretreated effluent (mechanical as well as chemical) showed better results than untreated effluent on some locations, but whether pretreatment is attractive is still subject to discussion. In the paper qualitative and quantitative results of experiments on several locations will be presented. An indication of operational costs will be given.

Large-scale application of UF and RO in the production of demineralized water

The water demineralization plant in Geleen (NL) for the production of boiler feed water from surface water with a maximum plant capacity of 1560 m 3/h uses reverse osmosis (RO) as the basic process step for desalination. In order to ensure maximum operational reliability of the RO, a dead-end ultrafiltration (UF) pretreatment step is used, giving the best feed quality with respect to RO fouling behaviour. In order to enable a proper design of the UF, a series of pilot-scale field experiments were carried out on-site. The overall process design, the experimental results of those experiments and the resulting selection of a suitable UF membrane are described. The subsequent selection of a suitable RO membrane for the process is then largely determined by economic aspects, which may be deduced from the membrane properties, representatively expressed in terms of salt passage and water flux (or required feed pressure) of the respective membrane.

Intermittent crossflushing of hollow fiber ultrafiltration systems

The successful operation of dead-end ultrafiltration systems depends on the effectiveness of cleaning (hydraulic and/or chemical cleaning). In this study, backwashing conditions such as pressure and duration were studied in order to maximize the net flux per cycle. Monitoring the backwash flux and turbidity of the backwash water showed that the efficiency of backwashing was more dependent on backwashing time than pressure. The effectiveness of backwashing was significantly improved by the use of crossflushing. In general, 99.9±1% flux restoration could be achieved when backwashing was preceded by crossflushing while 94.5±1% could be achieved by backwashing alone. Crossflushing was more effective when the crossflush velocity in the fiber was in the turbulent regime (v ≅ 1.6 m/s). Crossflushing and pressure pulsing can be employed to increase the efficiency of dead-end, hollow fiber ultrafiltration systems because the frequency of backwashing and chemical cleaning, permeate consumption and system down time are reduced.

Cross-flow filtration—an improved detergent removal technique for the preparation of liposomes

Using a cross-flow (tangential) filtration device, a lipid-detergent mixed micellar solution is filtered under pressure through a suitable membrane with a molecular weight exclusion limit of 10–50 kDa. In contrast to commonly used dead-end ultrafiltration filters, the solution flows tangential to the membrane surface. A wave-like flow of the solution over the membrane is effected due to the shape of the membrane, preventing clogging during filtration. Detergent monomers pass through the membrane and are collected as waste filtrate, whereas the mixed micelle solution is continuously pumped as retentate through the device until liposome formation is completed. Buffer solution is introduced into the system at the same rate as the filtrate is removed therefrom to keep the retentate volume constant. Variations of the following parameters were studied in detail: (1) membrane cutoff; (2) filtration area; (3) flow rate; (4) starting lipid/detergent ratio; (5) choice of detergent; and (6) cholesterol content. We determined the kinetics of detergent removal and compared filtration time, amount of filtrate, vesicle size and stability of the resulting liposome dispersions. Cross-flow filtration under optimum conditions results in liposome formation within minutes and in principle makes unlimited upscaling possible.

Effect of protein charge on cake properties in dead-end ultrafiltration of protein solutions

The properties of the filter cake formed on the retentive membrane in dead-end ultrafiltration of bovine serum albumin (BSA) solutions were obtained by using a batchwise filter which had a sudden reduction in its filtration area. Moreover, the zeta potential of BSA was measured by the electrophoretic light scattering method in order to investigate the correlation with the properties of the protein cake. The experimental data clearly demonstrated that the plots of such properties of the filter cake as the average porosity ɛav and the average specific filtration resistance αav against the absolute value of the zeta potential can almost be represented by a unique curve, respectively, irrespective of the sign of the zeta potential.

Separation of binary protein mixtures by ultrafiltration

Ultrafiltration of aqueous solutions of mixtures of bovine serum albumin(BSA) and egg white lysozyme was conducted using membranes which were almost completely retentive for BSA, but permeable for lysozyme, in order to investigate the factors influencing the filtration rate and the lysozyme rejection. The experimental data obtained in this study clearly demonstrate that the electrostatic interactions between dissimilar molecules may control the filtration rate and solute rejection in ultrafiltration of binary protein mixtures to a significant degree. It is also shown that increasing the shear stress acting on the membrane causes an increase in the lysozyme rejection, resulting in a reduced fractionation efficiency, and that ultrasonic irradiation is quite effective for fractionation of binary protein mixtures. Furthermore, the properties of the filter cake formed on the retentive membranes in downward dead-end ultrafiltration of binary protein mixtures have been studied in a batchwise filter which has a sudden reduction in its filtration area. In the pH range where both proteins are electropositive, the filter cake becomes loose and wet, thus reducing the flow resistance, while the cake becomes compact and dry in the pH range where the two proteins have opposite electrical charges.

Concentration and concentration gradient measurements in an ultrafiltration concentration polarization layer Part I: A laser-based refractometric experimental technique

In this paper we report on the development of an automated, laser-based refractometric technique for the measurement of concentration and concentration gradient in the concentration polarization layer during dead-end ultrafiltration. Spatially and temporally resolved measurements can be made to within 200 μm of the membrane. Solvent is pumped through the layer at a constant flow rate, and the driving pressure is also measured. The system was calibrated and tested using the biopolymer hyaluronan ( M n =434 000; M w =605 000; M z =827 000). Good repeatability, accuracy, and sensitivity were demonstated. These studies lay the foundation for a physico-chemical study of concentration polarization dynamics, reported in a companion paper.

Concentration and concentration gradient measurements in an ultrafiltration concentration polarization layer Part II: Application to hyaluronan

An optical refractometric technique has been used to acquire detailed spatially and temporally resolved concentration and concentration gradient measurements within an ultrafiltration concentration polarization layer. Solutions containing the polymer hyaluronan ( M n =434 000; M w =605 000; M z =827 000) in aqueous 10 mM NaCl and phosphate buffer solutions were added to a dead-end ultrafiltration flow cell and solvent was pumped through the layer at a constant flow rate. Membrane type (hydrophobic vs. hydrophilic track-etched polycarbonate), solvent type, initial polymer concentration, and pumping flow rate were changed systematically to quantify the effects of these parameters on the spatial and temporal development of the concentration polarization layer, and also on the pressure drop across the layer. The acquired data show discrepancies with theoretical predictions based on osmotic pressure and porous media modelling. These discrepancies may be due to a paucity of physico-chemical data required in the modelling, or the presence of mechanical stresses within the polarized layer not taken into account in the theory. An improved approach for fitting polarized layer concentration profiles is described.

Protein ultrafiltration concentration polarization layer flux resistance I. Importance of protein layer morphology on flux decline with gelatin

The effect of added ethanol and (NH 4) 2SO 4 on the morphology of the gelatin concentration polarization (CP) layer during ultrafiltration (UF) was analyzed using protein ternary phase diagrams, as well as turbidity and intrinsic viscosity measurements. CP layer resistance was characterized by measuring the flux decline index (FDI) of a 0.10 wt% gelatin solution in a dead-end UF system at 40°C. Ethanol concentrations up to 30 wt% produced an aggregated CP layer morphology with a moderate FDI. However, above 30 wt%, a swollen macromolecular gel formed on the membrane surface, resulting in a lower FDI. Low salt levels led to the formation of compact protein particles in the CP layer that had a moderate FDI. In contrast, a concentration of 8 wt% produced a high FDI coacervate structure. The highest FDI was observed at salt concentrations above 12 wt%, where the surface morphology consisted of a dense particulate gel. Overall, the FDI could be altered by a factor of ∼3 by manipulating the concentration of solute in the solution.

Electrical conductivity measurement of filter cake in dead-end ultrafiltration of protein solution.

The dynamic ultrafiltration process of protein solution is experimentally investigated by measuring electrical conductivity within the filter cake. Three pairs of platinum wire electrodes were employed to measure the formation process of the filter cake on the membrane surface. The filter cakes formed from protein solutes behave as highly compressible cakes and shifting of pH from 5 to 4 or 9 results in a decrease in cake compressibility, as well as specific filtration resistance. The experimental results for bovine serum albumin and soy protein solutions coincide favorably with the calculated results based on the compressible cake filtration theory.

Obtaining the concentration-dependent diffusion coefficient from ultrafiltration experiments: Application to hyaluronate

A method for determining the concentration-dependent mutual diffusion coefficient D(C) of a macromolecule-solvent combination over a wide macromolecular concentration range is presented. All necessary data are gathered from a single experiment, in which polymer concentration profiles are measured during one-dimensional dead-end ultrafiltration. Based on these profiles, the convection-diffusion equation is used to deduce the dependence of D on C. To demonstrate the utility of the approach, studies using sodium hyaluronate dissolved in either 10 mM NaCl or a phosphate buffer were carried out. For hyaluronate (HA) in 10 mM NaCl, the mutual diffusion coefficient varies approximately linearly with concentration according to D(C) = 4.1 × 10−6 C0.96 cm2/s in the range 0 ≤ C ≤ 0.6 mass %, for C expressed in mass %. However, transition points (slope changes) in the D(C) curve are present at C ≅ 0.7 mass % and C ≅ 1.4 mass %. For HA in the phosphate buffer, the mutual diffusion coefficient is well described by D(C) = 1.9 × 10−6 C0.825, for 0 ≤ C ≤ 1.8 mass %. These values agree well with previously published data. The technique is robust, and permits reasonably high polymer concentrations to be easily studied. © 1996 John Wiley & Sons, Inc.

5536872 Michael addition process: Sabahi Mahmood; Lilje Kenneth C, Baton Rouge, LA, United States assigned to Albemarle Corporation

Ultrafiltration (UF) is considered as a suitable treatment process after conventional wastewater treatment to produce reuse water. Nevertheless, fouling affects the performance of UF to a large extent. As biopolymers (mostly macro polysaccharide-like and protein-like molecules) have been identified as major foulants affecting the filterability of water in dead-end UF, the present study focuses on investigating the reversibility of biopolymer fouling occurring at different biopolymer mass loads to the membrane and under different compression conditions. UF-membrane stirred cell tests using five cycles show that filtering treated domestic wastewater leads to a significant permeability reduction due to the accumulation of biopolymers on the membrane surface and/or in the membrane pores. Although they can be removed by hydraulic backwashing, an increased mass load of biopolymers reduces the removal efficiency. This correlation was verified using a UF pilot plant filtering treated wastewater (secondary effluent or slow sand filtrate). The effect of biopolymer fouling layer deformation on its reversibility was studied using multi-cycle membrane filtration tests under different filtration pressures. The results showed that higher filtration pressures result in more compact biopolymer fouling which is more difficult to be hydraulically backwashed. This phenomenon was also confirmed by pilot-scale UF experiments.

Unstirred dead-end ultrafiltration: a method to determine diffusion coefficient or osmotic pressure for non-charged macromolecular solutions

Concentration polarization phenomenon in ultrafiltration or reverse osmosis processes is governed by a non linear convective-diffusion equation. General unsteady state equations associated with the osmotic theory relationship are solved numerically. Unstirred ultrafiltration flux declines are accurately described when the concentration dependence of the osmotic pressure or diffusion coefficients are known. Inversely, when the experimental evolution of the permeate flux is introduced in the calculation, it is possible to determine osmotic pressure or diffusion coefficients as a function of solute concentration with a good precision. The results obtained for four different kinds of well-defined non charged polymers agreed well with available literature data.

Upward Dead-End Ultrafiltration of Binary Protein Mixtures

Abstract Upward dead-end ultrafiltration of aqueous solutions of mixtures of bovine serum albumin (BSA) and egg white lysozyme was conducted using membranes which are almost completely retentive for BSA but permeable for lysozyme. The dependence of lysozyme rejection and the filtration flux rate on pH and the addition of salts has been investigated. The experimental data obtained in this study clearly demonstrate that the electrostatic interactions between dissimilar molecules may control the solute rejection and the filtration rate in upward ultrafiltration of binary protein mixtures. For instance, the BSA and lysozyme molecules have opposite electric charges at pH 7. Consequently, lysozyme rejection is large because both molecules within the filter cake pack more tightly due to heterocoagulation. On the other hand, the BSA and lysozyme molecules have the same electric charge at pH 4. Thus, lysozyme is rejected by the filter cake of the retained BSA molecules due to repulsive electrostatic interactions b...

Clarification of Green Tea Extract by Microfiltration and Ultrafiltration

Membrane clarification of green tea extract was studied as a treatment to reduce sediments in packaged drinks and as a pretreatment for concentration processes. The flux and variation of components were examined in dead-end and crossflow filtration with several types of membranes. In dead-end ultrafiltration, the flux reduction rate was small, although the initial flux was similar to the final flux in microfiltration. Prefiltration was effective in decreasing the reduction rate of flux. As the pore size of microfiltration membranes became smaller, the dry weight decreased gradually and the optical transmission at 660 nm increased. By ultrafiltration, 30–50% pectin, 3–11% catechins and, 7–20% caffeine were rejected. Crossflow filtration was effective in keeping the flux high. The ultrafiltration spiral membrane (pore size: 0.008 μm) was selected for repeated batch clarification of prefiltered green tea crude extract and showed reproducible performance.