Search for: All records

Ammonia present in many industrial process streams and effluent streams is beginning to be recovered by means of microporous hydrophobic hollow fiber-based membrane contactor devices with gas-filled pores; the process is often characterized as supported gas membrane (SGM) process. Ammonium sulfate is usually obtained in a sulfuric acid stream on the other side of the membrane. It is useful to develop a quantitative basis for the extent of ammonia removal in such devices. Unlike deoxygenation of aqueous streams in such devices, membrane resistance is quite important for ammonia transport. Ammonia transport modeling in such devices is hampered by the complexity of feed liquid flow in the shell side of commercially used devices and lack of information on membrane resistance where membrane tortuosity introduces considerable uncertainty. The approach adopted here involves studying ammonia transport with the feed solution flowing through the hollow fiber bore where the fluid mechanics is simpler than shell-side flows. Comparison of model-based predictions of overall mass transfer coefficient (ko) with experimentally observed values allows estimation of the membrane mass transfer coefficient (km). One can use such estimates of km to model the observed ammonia transport in small crossflow devices and develop an empirical guidance of the dependences of the shell side mass transfer correlations. Guided by such information and deoxygenation SGM literature, a model was developed for large modules used for ammonia recovery via SGM. Model predictions of performances of the large modules are likely to be useful for various process considerations including the effect of temperature and feed flow rate variations on ammonia removal. 

To unravel fouling and defouling mechanisms of protein, saccharides and natural organic matters (NOM) on polymeric membrane during filtration, this study investigated filtration characteristics on polyvinyl chloride (PVC) ultrafiltration membranes with bovine serum albumin, dextran, humic acid as model foulants. Membrane fouling and defouling performances were analyzed through monitoring the flux decline during filtration and flux recovery during physical backwash. Physico-chemical properties (e.g., hydrophobicity and surface charge) of PVC membrane and foulants were characterized, which were used in the extended Derjaguin–Landau–Verwey–Overbeek (EDLVO) theory to calculate the interaction energies between membrane foulant and foulant-foulant. The results showed that at the later filtration stages the fouling rate was strongly correlated with the deposition rate, which was determined by the interaction energy profile calculated by EDLVO. Moreover, the adhesion forces of membrane–foulant and foulant–foulant were further measured by atomic force microscopy (AFM) with modified colloidal probes. A positive correlation (R2 =0.845) between particle detachment rate (determined by adhesion force) and defouling rate was developed for BSA and HA foulants that led to cake layer formation. By contrast, dextran defouling rate was off this correlation as dextran partially clogged membrane pores due to its smaller size. 

Organic solvent filtration is an important industrial process. It is widely used in pharmaceutical manufacturing, chemical processing industry, semiconductor industry, auto assembly etc. Most of the particle filtration studies reported in open literature dealt with aqueous suspension medium. The current work has initiated a study of cross-flow solvent filtration behavior of microporous ethylene chlorotrifluoroethylene (ECTFE) membranes using 12 nm silica nanoparticles suspended in an aqueous solution containing 25% ethanol. In the constant pressure mode of operation of cross-flow microfiltration (MF), permeate samples were collected at different time intervals. The permeate particle size distribution (PSD) results for different experiments were identical. Particle agglomerates having less than 100 nm size can pass through the membrane; some fouling was observed. The governing fouling mechanisms for tests operated using 3.8×10−3 kg/m3 (3.8 ppm) at 6.9×103 Pag and 1.4×104 Pag were pore blocking. For tests conducted using 3.8×10−3 kg/m3 (3.8 ppm) at 27.6×103 Pag (4 psig) and 1.9×10−3 kg/m3 (1.9 ppm) at 6.9×103, 13.8×103 and 27.6×103 Pag (1, 2 and 4 psig), the mechanism was membrane resistance control. Less particles got embedded in membrane pores in experiments operated using suspensions with lower or higher particle concentrations with a higher transmembrane pressure. This is in good agreement with the values of the shear rate in the pore flow and scanning electron microscope images of the membrane after MF. In the dead-end mode of operation of solvent filtration using methanol, ethanol and 2-propanol, the permeate flux behavior follows Jmethanol > Jethanol > J2-propanol at all testing pressures. The values of permeance (kg/m2-s-Pa) determined from the slope of the linear plot of filtration flux vs. the applied pressure difference across the membrane, were 3.9×10−4, 2.3×10−4 and 3.0×10−5 for methanol, ethanol and 2-propanol, respectively. Further exploration was made on solvent sorption results reported earlier. The critical temperature of selected solvents shows a better correlation with solvent sorption rather than the solubility parameter.