(727a) Membrane Mineral Scaling in Semi-Batch and Steady State Reverse Osmosis Desalination - a Comparative Study

RO membrane desalination has proven to be a robust technology for desalting both brackish water and seawater. However, despite the popularity of RO technology, membrane mineral scaling remains a major impediment to achieving high recovery, particularly in brackish water desalination, and also leads to increasing operational cost (due to required membrane cleaning cleaning) and shortening membrane lifetime. Recent studies have proposed the use of semi-batch RO as an alternative operational approach that could enable high recovery operation at reduced level of antiscalant dosing which would provide more effective means of mitigating mineral scaling. In semi-batch RO operation, continuous concentrate recycling is adopted in a filtration period to enhance the system water recovery. During the filtration period salt concentration (including the concentration of sparingly soluble salts) rapidly increases in the feed flow channel as salt accumulates in the feed flow channel. Once the osmotic pressure in the feed channel reaches the upper allowed limit, the concentrate holdup is discharged and can be flushed with the raw-feed water. Such intermittent flushing after each filtration period has been said to reduce membrane mineral scaling by resetting the nucleation induction period (or “induction clock”) for the onset of mineral scaling on the membrane surface. Current studies have argued that the above operational mode would lead to more effective RO operation with respect to scale mitigation relative to conventional steady state RO. In order to assess the above hypothesis, an investigation was undertaken to evaluate the impact of repeated filtration-flushing cycles in semi-batch RO operation on mineral scaling. In order to unambiguously assess mineral scaling under semi-batch operation, a direct real-time membrane surface imaging system was developed to detect the onset of membrane scaling, with gypsum (calcium sulfate dihydride) selected as the model mineral scalant. The surface monitoring system was interfaced with a spiral-wound pilot scale RO system, capable of achieving up to 90% water recovery. RO desalination tests were then carried out in both semi-batch and steady-state operational modes to compare the above two operational modes with respect to resetting of the scaling induction period and quantify mineral scale kinetics at various water recoveries. Real-time membrane surface image analysis enabled quantification of gypsum scaling kinetics in terms of both the number density of gypsum crystals and the extent of membrane surface scale coverage. The above information enabled evaluation of: (a) impact of intermittent raw-feed water flushing for semi-batch RO operation on the formation of nuclei and rate of crystal growth, (b) effect of convective residence time in the RO system on the scale induction period, (c) mineral scaling at different levels of raw feed supersaturation with respect to gypsum, and (d) mineral scaling at semi-batch RO ad steady state RO at comparable levels of mineral saturation in the RO feed channel. Given the ability for early detection of mineral scaling, via the current direct and real-time membrane surface imaging, the present work provides an unambiguous demonstration that, contrary to hopes raised regarding the superiority of semi-batch RO, mineral scaling can be significantly exacerbated in this operational mode relative to steady state RO operation. However, there are certain limited operational domains in which semi-batch RO may be advantageous. The fundamental results of the current study provide a basis for assessing the recovery limits that are imposed by mineral scaling and possible operational approach to scale mitigation for semi-batch and steady-state RO.