| Summary: | Membrane filtration systems are successful methods to make high-quality water. An important issue and most challenging problem regarding these systems is fouling which arises from particle deposition in and on the membrane surface. Membrane fouling reduces performance of membranes and increases operational costs. Due to this problem, a reliable index to predict the fouling potential of feed water is important in preventing and characterizing membrane fouling and for monitoring the performance of pre-treatment during plant operation. The conventional fouling indices such, as SDI and MFI, are usually determined via off-line measurement techniques based on dead-end filtration, which does not consider the actual flow conditions. To have a better understanding of the conventional fouling indices, the effect of several parameters on SDI, MFI, MFI-UF (constant pressure and constant flux) were investigated. In contrast with MFI, the SDI results confirmed the non-linear response to foulants concentration. The specific cake resistance for colloidal silica and humic acid was found to be maximum at low concentrations and gradually declined at higher concentrations. It was also highlighted that fouling by mixed foulants cannot be related simply to the individual MFIs. The effect of salt concentration on MFI-UF (constant pressure and constant flux) was also investigated over a broad range of NaCl (0 to 25,000 mg/L) for colloidal silica and humic acid. The results indicated that MFI-UFconst.pressure of the colloidal silica passed through a maximum (under constant pressure filtration) while for humic acid MFI-UFconst.pressure increased monotonically. To overcome the limitations of these techniques, an on-line flow simulator, the feed fouling monitor (FFM), has been developed for characterization of the fouling propensity of feed water to RO. The FFM is based on the principle of particle deposition and cake layer formation on a small ‘collection’ membrane in the device under cross flow conditions, which can be correlated to the particulate fouling (including colloidal particles and macromolecules) encountered in a RO operation. The information obtained from FFM was used to estimate cake resistance in RO in order to predict RO fouling profiles at the same hydrodynamic conditions. However, the predicted RO profiles were found to underestimate the actual RO performance due to ignoring the contribution of the cake-enhanced osmotic pressure (CEOP) effect. The model was extended to include the contribution of CEOP and ultrasonic time domain reflectometry (UTDR) was used to monitor the rate of cake thickness increase in RO in order to estimate the cake-enhanced osmotic pressure (CEOP). The cake resistance (obtained from FFM) combined with CEOP could more accurately predict the RO fouling profiles using colloidal silica as a foulant model. A salt tracer response technique (STRT) was also used to measure directly concentration polarization trends and estimate the contribution of CEOP under constant flux using both organic and inorganic foulant models. The predicted RO fouling profiles obtained from combination of FFM and STRT provided a good estimation of the transmembrane pressure (TMP) rise during RO fouling for both humic acid and colloidal silica. The contribution of CEOP to TMP rise was typically greater than the cake resistance.
|
|---|