Optimization of Hollow-Fiber Design and Low-Pressure Membrane System Operation

An approach is presented for the numerical optimization of low-pressure membrane filtration processes. A multidimensional optimization of an ultrafiltration system is formulated for cost minimization and numerically solved for key optimal design and operating variables. Hollow-fiber ultrafiltration operation under steady-state conditions is assumed and optimized with respect to fiber radius, fiber length, crossflow velocity, transmembrane pressure, and system recovery. Optimizations are performed over variable raw water conditions using a sequential quadratic programming (SQP) algorithm. For typical small to moderately sized low-pressure membrane facilities (≈1 mgd), optimal fiber design and membrane system operation is predicted to be largely influenced by the characteristic dominance of capital costs over operating costs. Thus, total treatment costs tend to be optimal at values of decision variables where permeate fluxes are maximized, within the constraints prescribed by the system, and assuming a fixed membrane cost per unit area. For raw waters demonstrating apparent to significant membrane fouling, optimal membrane treatment is predicted to be achieved by using relatively narrow hollow fibers and relatively high crossflow velocities. For relatively clean raw waters demonstrating very high sustainable permeate fluxes, operating at low crossflow velocities—or perhaps even under the dead-end mode of operation—appears to provide the most cost-effective operation.     

NOM Accumulation at NF Membrane Surface: Impact of Chemistry and Shear

The effects of solution chemistry, surface shear, and composition of natural organic matter (NOM) were investigated for their impact on accumulation of foulant material at the surface of charged polymeric nanofiltration membranes. The source of NOM was the Suwannee River. A bench-scale, batch recycle system was used with 20 hollow fiber, nanofiltration membranes. Membrane flux decline and foulant accumulation increased at low pH and high ionic strength as a result of neutralization of charge, electric double layer compression, and the apparent shift in conformation of charged NOM macromolecules. The rate of NOM accumulation decreased with operating time, suggestive of an eventual steady state between adsorption and desorption. The effect of NOM composition on membrane fouling could not be discerned by a standard technique to isolate hydrophobic and hydrophilic NOM fractions, quite possibly because of the fractionation methodology's failure to recover a small but important fouling fraction or because of NOM interactions that are lost when individual fractions are separately tested. However, a greater percentage of the hydrophilic than hydrophobic fraction permeated the membrane, in agreement with prior observations by others. Increasing the cross flow velocity from 85 to 255 cm∕s reduced the extent of flux decline, presumably due to hydrodynamic disruption of cake layer formation.     

Modeling Formation of Natural Organic Matter Fouling Layers on Ultrafiltration Membranes

Three simple mathematical models to describe fouling of an ultrafiltration membrane by natural organic matter (NOM) are developed and compared. These models attribute the fouling to: (1) an increase of the effective pore length by an amount equal to the thickness of the NOM gel layer that forms on the membrane surface; (2) formation of a uniform, microporous NOM gel layer on the membrane surface, made of primary particles comprising tens to hundreds of NOM molecules; or (3) narrowing of the membrane pores by sorption of a monolayer of NOM molecules along the full length of each pore. The key parameters characterizing each model are identified and estimated based on data for flux and film growth gathered in the same system. In each case, the estimated parameter values are plausible in light of the known physical properties of the membrane and NOM molecules.     

Determination of Perchlorate Rejection and Associated Inorganic Fouling (Scaling) for Reverse Osmosis and Nanofiltration Membranes under Various Operating Conditions

This study focused on perchlorate (ClO4?) rejection and flux-decline in bench-scale cross-flow flat-sheet filtration for two reverse osmosis (RO) and two nanofiltration (NF) membranes with a natural water, and addressed estimation of precipitative fouling/scaling with inorganic salts and characterizations of inorganic fouling and antiscalants. Thus the study considered tradeoffs between productivity (increased recovery and flux) versus ClO4? rejection versus membrane fouling/scaling. In this study, the rejection of water quality parameters (cations, anions, dissolved organic carbon, UVA254, total dissolved solids) and flux-decline trends for four different membranes were investigated over a various range of operating conditions (i.e., J0/k ratio and recovery). Inorganic foulants on the membrane surface were analyzed by various methods (i.e., x-ray diffraction and scanning electron microscopy), and demonstrated inhibition effects of antiscalant. With increasing recovery and J0/k ratio, high productivity (flux) was achieved, however, the rejections of perchlorate and other water quality parameters decreased and the precipitative fouling/scaling potential of membranes increased. At the same operating conditions in the presence of an antiscalant, embodying phosphonate functional groups, flux decline trends for the four membranes indicated lower scale formation supported by the results of the fouled membrane characterizations.     

Membrane Fouling in Membrane Bioreactors for Wastewater Treatment

Membrane bioreactors (MBRs), in which membranes are applied to biological wastewater treatment for biomass separation, provide many advantages over conventional treatment. However, membrane fouling in MBRs restricts their widespread application because it reduces productivity and increases maintenance and operating costs. Recently much research and development has taken place to investigate, model, and control membrane fouling processes. However, unified and well-structured theories on membrane fouling are not currently available because of the complexity of the biomass matrix, which is highly heterogeneous and includes living microorganisms. Membrane fouling in MBR systems can be reversible (i.e., removable by physical washing) or irreversible (removable by chemical cleaning only), and can take place on the membrane surface or into the membrane pores. Although establishing a general model to describe membrane fouling in such a process is made extremely difficult by the inherent heterogeneity of the system, the nature and extent of fouling in MBRs is strongly influenced by three factors: biomass characteristics, operating conditions, and membrane characteristics. Fouling control techniques which have been investigated include low-flux operation, high-shear slug flow aeration in submerged configuration, periodical air or permeate backflushing, intermittent suction operation or addition of powdered activated carbon (PAC). Of these, only PAC addition is currently not used in existing large-scale installations.     

Pore Blockage Effects on Atrazine Adsorption in a Powdered Activated Carbon/Membrane System. II: Model Verification and Application

The bisolute model developed in Part I of this study to describe the effect of pore blockage on trace organic compound adsorption was verified and then used to study the roles of various design and operating variables on process performance efficiency. Continuous flow experiments were conducted with atrazine and a pore-blocking macroelectrolyte, poly(styrene sulfonate) (PSS-1.8k) using two powdered activated carbons (PACs) in a bench-scale PAC/microfiltration (MF) system. The model was calibrated with one set of experimental data, and verified with additional data obtained at different operating conditions for both PACs. PAC B, which has a large mesopore surface area and volume, was found to be less affected by pore blockage at a given PSS-1.8k surface loading compared to PAC A, which has relatively small mesopore surface area and volume. However, it was also shown that when PAC B was fully loaded with PSS-1.8k, the pore blockage effect on the rate of atrazine diffusion was even greater than that for PAC A due to the higher PSS-1.8k adsorption capacity of PAC B. Model simulations were conducted to investigate the effects of reactor hydraulic retention time, membrane cleaning interval, influent PSS-1.8k concentration, PAC dosage, PAC dosing scenario, and the quality of membrane cleaning water on system performance. Optimal PAC/MF system operating conditions were also determined based on model simulations. The study results showed the effects that different concentrations and adsorbent loadings with pore blocking compounds, such as the pore blocking fractions of natural organic matter, can have, and the importance of mesopores in mitigating the detrimental effect of pore blockage.     

Modeling Manganese Oxidation with KMnO4 for Drinking Water Treatment

Manganese (Mn) in a drinking water distribution system can cause multiple aesthetic problems including discolored water and fouling or scaling of fixtures. Oxidation and solid-liquid separation processes are typically employed at a treatment plant to limit the concentration of Mn entering the distribution system. Potassium permanganate (KMnO4) is commonly used to oxidize the manganous ion (Mn+2) to manganese oxide (MnO2). In this study, a mechanistic model is applied to the oxidation of manganese at a treatment plant. Literature kinetic constants (determined with artificial water) are compared with the values obtained for the plant's natural water. The solution and surface phase oxidation rate constants determined with the natural water are two to six orders of magnitude less than those determined with the artificial water. The reduced oxidation rate in the natural water is attributed to the presence of dissolved organic matter, which can exert a competitive demand on the oxidant and interfere with the oxidation by complexing Mn+2. The development of an additional rate constant for the oxidation of dissolved organic matter improves the modeling results for KMnO4 concentration versus time, but only marginally explains the Mn+2 oxidation rate differences.     

Impacts of Hydrophilic Membrane Additives on the Ultrafiltration of River Water

One of the most serious disadvantages of membrane applications in water treatment is the decreasing water permeation rate with time, which is often called fouling. This study investigates surface modification of polyethersulfone (PES) ultrafiltration membranes as a fouling reduction strategy for drinking water treatment applications. Surface modification was achieved through the addition of three different tailor-made hydrophilic surface modifying macromolecules (LSMM200, LSMM400, and LSMM600). Flat sheet membranes were prepared via a single-step casting procedure; their surface hydrophilicity was quantified via contact angle measurements. The incorporation of hydrophilic additives produced slightly more hydrophilic membranes (contact angle reduction of up to 8°) and improved membrane performance compared with the PES membrane without blending. In the treatment of highly colored river water, LSMM400- and LSMM600-modified membranes achieved up to 32% higher final fluxes. Surface modification resulted in significantly decreased flux reductions and natural organic matter accumulation. Dissolved organic carbon removals were approximately 70% for all the membranes studied. No clear correlation between membrane hydrophilicity and fouling reduction was observed.     

Process Limits of Municipal Wastewater Treatment with the Submerged Membrane Bioreactor

The submerged membrane bioreactor (SMBR) is a promising technology for wastewater treatment and water reclamation. This paper presents results from two pilot scale SMBR systems operating in parallel on municipal wastewater in San Diego, Calif. The SMBRs were operated to address the limitations and advantages of the SMBR process compared to conventional activated sludge processes. Minimal membrane fouling was observed throughout the year of testing with the exception of the process limitations. Both pilot units provided consistently high quality effluents throughout the study, even when operating at hydraulic retention times as low as 1.5 h. Two sets of experiments were conducted to identify different fouling conditions. The first experiments were conducted to explore operation at high suspended solids concentrations. The SMBR process experienced adverse performance at mixed liquor suspended solids concentrations greater than approximately 20?g/L. The second experiments explored operation at low mean cell residence time (MCRT). At an MCRT of <2 days, membrane fouling was rapid. Chemical cleaning with sodium hypochlorite solution provided full recovery of the membrane permeability.     

Morphology of Particle Deposits

The premise that the morphology of deposits formed from colloids, aerosols, and other particulate materials can be predicted from their transport properties and surface chemistries is explored. Universality classes that group the conditions of particles in suspension to the structure of a deposit composed of a large number of particles are considered in the context of their importance in natural and engineered systems encountered in environmental engineering. Evidence for one such “transport-morphology link” is presented, and the implications of this linkage in filter modeling, prediction of sediment characteristics, and membrane fouling are presented.     

Applicability Assessment of Subcritical Flux Operation in Crossflow Microfiltration with a Concentration Polarization Model

In the process of crossflow microfiltration, a deposit of cake layer tends to form on the membrane, which usually controls the performance of filtration. However it is found that there exists a condition under which no deposit of cake layer is made. This condition is called the subcritical flux condition and the maximum flux in the condition here is called the critical flux. Which means, it is a flux below which a decline of flux with time due to the deposit of cake layer does not occur. This study develops a concentration polarization model to predict the critical flux condition and to study about its characteristics. The model is verified with experimental results. For the model, the concept of effective particle diameter is introduced to find a representative size of various particles in relation to diffusive properties of particles. The modeling and the experimental results include that the critical flux condition can be determined by the use of effective particle diameter and the ratio of initial permeate flux to crossflow velocity. This study also finds that the (sub)critical flux operation is limited for the real world application because of the limitation to increasing crossflow velocity and its sensitivity to the change of feed composition.     

Removal of 17β Estradiol and Fluoranthene by Nanofiltration and Ultrafiltration

With the recent emergence of endocrine disrupting compounds as an important potable drinking water and reclaimed wastewater quality issue, the removal of two estrogenic compounds (17β-estradiol and fluoranthene) by nanofiltration and ultrafiltration membranes was investigated. A less hydrophobic organic compound model species [parachlorobenzoic acid (PCBA)] was tested. 17β-estradiol (E2), fluoranthene, and PCBA were applied to the membrane in the presence and absence of natural organic matter (NOM). Both batch adsorption and dead-end stirred-cell filtration experiments indicated that adsorption is an important mechanism for transport/removal of relatively hydrophobic compounds, and is related to the octanol-water partition coefficient (KOW) values. All filtration measurements were performed approximately the same permeate flow rate in order to minimize artifacts from concentration polarization varied with different hydrodynamic operating conditions at the membrane interface. The percent removal by dead-end stirred-cell filtration ranged from 10 to >95% depending upon membrane pore size/hydrophobicity and presence/absence of NOM at an initial concentration ranging from 0.1 to 0.5 μM. Additional batch adsorption experiments with radio-label (3H) E2 at lower concentrations ranging 0.025 to 5 nM showed that E2 removal due to adsorption was independent of its initial concentration. Adsorption occurs both on the membrane surface and interior membrane pore surfaces. However, adsorption was insignificant for PCBA (log?KOW = 2.7), but removal presumably occurred due to electrostatic exclusion. Partition coefficients (log?K) of 0.44 to 4.86 measured in this study increased with log?KOW and membrane pore size.     

Estimating the Removal of Anthropogenic Organic Chemicals from Raw Drinking Water by Coagulation Flocculation

A mathematical model was developed to estimate the efficacy of coagulation–flocculation treatment for removing neutral hydrophobic organic chemicals from raw drinking water. The model assumed that the only significant removal mechanism was the destabilization and settling of organic matter containing sorbed anthropogenic organic compounds. The model was validated with standard jar tests using compounds with a range of hydrophobicities (log?Kow = 1.89?to?5.48), including contaminant candidate list chemicals, pesticides, pharmaceuticals, and endocrine disrupting chemicals. Final concentrations of test compounds after coagulation and flocculation were in good agreement with model estimations for synthetic waters composed of Aldrich (Milwaukee, WI) humic acid solutions. The final compound concentrations in coagulated natural waters from two drinking water reservoirs were about 80% lower than those estimated with the model. Overestimations of treated water concentrations by the model were attributed to an increase in sorption by natural organic matter when coiled in aluminum hydroxide flocs, compared to sorption to dispersed natural organic matter in untreated water.     

Existence of Critical Recovery and Impacts of Operational Mode on Potable Water Microfiltration

Results from a potable water microfiltration (MF) pilot study employing untreated surface water are reported. The effects of filtrate flux and recovery on direct flow, outside-inside, hollow fiber MF fouling rates, and backwash effectiveness are presented. Constant flux experiments suggested the existence of a critical recovery below which MF fouling rates were low and effectiveness of backwashes was high and relatively independent of the recovery. However, in the range of experimental conditions investigated, fouling rates increased dramatically and backwash effectiveness decreased steeply when this critical recovery was exceeded regardless of the flux. In general, for a fixed recovery, specific flux profiles analyzed on the basis of volume filtered per unit membrane area were insensitive to filtrate flux. Fouling was accelerated by operating membranes at constant flux rather than at constant pressure, in part, because of membrane compaction and cake compression. Changing the mode of filtration between constant flux and constant pressure is shown to have no effect on MF filtrate water quality. For any given capacity, membrane area requirements are decreased, and power requirements are increased when membranes are operated at constant flux rather than at constant pressure.     

Neural Network Modeling of Organics Removal by Activated Carbon Cloths

The adsorption of organic compounds onto an activated carbon cloth is studied in a dynamic reactor. An experimental design is carried out to investigate the influence of operating conditions (initial concentration C0, flow velocity U0, and bed thickness Z) and adsorbate's characteristics. A slow intraparticular diffusion is shown by flattened breakthrough curves, and adsorption capacities are high and range between 50 and 250 mg g?1. The transfer zone Z0, assessed by the Adams and Bohart equation, is low (3 mm). All experimental results are modeled by a neural network to take into account the specific diffusion of cloths. Parameters related to the adsorbate-adsorbent affinity in a batch reactor are consequently introduced in the input layer of the neural network (intraparticular coefficient Kw and Freundlich parameters Kf and l∕n), added to operating conditions whose influence was shown (C0, U0, and Z) and time t. The statistical quality of the neural network modeling is high (r2 = 0.956). Furthermore, the Garson connection weight method allows the relative influence of input neurons to be determined. This analysis confirms the influence of parameters relative to adsorbant-adsorbate affinity.     

Use of Membrane Bioreactor for Biodegradation of MTBE in Contaminated Water

An ultrafiltation membrane bioreactor was evaluated for biodegradation of methyl tert-butyl ether (MTBE) in contaminated water. The system was fed 5 mg/L MTBE in granular activated carbon (GAC) treated Cincinnati tap water containing ample buffer and nutrients. Within 120 days the culture had adapted to membrane operational conditions and was consistently achieving greater than 99.95% biological removal of both MTBE and tert-butyl alcohol. This condition was steadily maintained for the next 200 days of study. Effluent dissolved organic carbon values remained at or below concentrations of the feed GAC treated tap water alone. An increase in biomass concentration as measured by volatile suspended solids was observed to correlate with an increase in MTBE removal efficiency. Some operational observations, including fouling, recovery from an accident, and overall performance, are described.     

Fluorescence Technique for Rapid Identification of DOM Fractions

The dissolved organic carbon parameter has typically been used as a measure of organic content in natural water. However, dissolved organic carbon is an aggregate parameter and does not provide information on the organic character of natural organic matter in water. Natural organic matter from New Jersey surface water sources was isolated and fractionated by resin adsorption into hydrophobic acid, hydrophobic neutral, hydrophobic base, hydrophilic acid, hydrophilic neutral, and hydrophilic base. The spectral fluorescent signature technique was developed for the identification of the six dissolved organic matter (DOM) fractions through a multiple linear regression model. High sensitivity, rapid identification, and quantification of DOM fractions are among the main advantages of this technique. The technique∕model has spatial and temporal potential use for the rapid qualitative and quantitative measurement of the problematic DOM fraction(s) for source water characterization∕assessment and water treatment process optimization.     

Autologous hematopoietic stem cell transplantation in chronic myeloid leukemia

A circuit that simulates T-type calcium-channel current characteristics of the sinoatrial (SA) node was developed from discrete electronic components and tested at physiologic membrane voltage ranges. The circuit design was based on the T-type calcium-channel current dynamics obtained from a mathematical model of the SA node membrane, which, in turn, is based on physiologic data. The design was held at a resting membrane potential and then stepped to new voltages over the entire operating range of the T-type calcium channel. The circuit was validated by comparing its transient response current with the predicted current from the mathematical model. In addition, the peak currents of the circuit were compared with plots of peak current obtained from the mathematical model and physiologic data. By showing that the electronic circuit mimics the T-type calcium-channel current dynamics found within the SA node, the results may provide a foundation for developing a novel cardiac pacemaker that is based on the ion-channel characteristics of excitable tissue.     

Dynamic optimization analysis for equipment setup problems in endurance cycling

The goals of the work reported by this article are two-fold. The first is to develop a dynamic optimization framework for analysis of equipment setup problems in endurance cycling. The second is to illustrate the application of the approach by determining an optimal chainring shape. To achieve these goals, a mathematical model of the pedaling motion for given trajectories of the net joint moments and the rate of change of the chainring radius was derived, and chainring optimization was posed as an optimal control problem. The cost functional produced a chainring shape that reduced the cost of endurance cycling at 250 W and 90 rpm, apparently by taking advantage of mechanical interactions that arise as a natural consequence of the movement. However, the predicted joint moments required larger peak values during phases of significantly increased joint velocity. Thus, the 'optimal' performance predicted by the cost functional appears opposed to expectations based on muscle mechanics and illustrates the need for further analysis of endurance cycling with a physiologically based cost functional.     

Membrane Fouling Test: Apparatus Evaluation

Flux decline with time is one of the most serious shortcomings of microfiltration and ultrafiltration membranes. It is highly desirable to have a membrane (fouling) testing procedure that is short in duration, utilizes a minimum amount of test solution, only requires a small membrane area, and is representative of the large-scale process. The objective of this study was to compare the results of the testing of a given membrane using a number of different test units (reverse osmosis, ultrafiltration, dead-end, and cross-flow cells) and testing procedures. It was of particular interest to determine if smaller cells used in the literature perform similarly to the Sepa CF cell, as it is a standard. During six-day runs the flux decline of the polyethersulfone membrane tested was mainly caused by membrane compaction and much less due to fouling. As various membrane materials compact to a different extent, studies into the fouling characteristics of different types of membranes should incorporate precompaction and pure water testing to quantify the contribution of membrane compaction and true fouling to the overall flux decline. The dead-end cell performed very differently from continuous cells, so their use is not recommended. The six-day continuous flow tests showed that the reverse osmosis (RO), ultrafiltration (UF), and cross-flow (CF) cells yielded very similar dissolved organic carbon removals and flux decline, despite UF and RO cells using membrane coupons eight times smaller than CF cells.