In flat sheet submerged membrane, membrane panels are stacked against each other with 5-7 mm spaces in between two panels. All the bubbles released from the diffusers located underneath the membrane panels rise through the in-between spaces while physically contacting with membrane surface. Large bubbles are likely more effective in membrane scouring due to the strong falling film and wake effects, but a large amount of bubbles are required to cover the entire membrane surface. In contrast, small bubbles generate weaker falling film and wake effects, but they can cover the entire membrane surface at lower air flow rate. The optimum bubble size must be decided considering the membrane scouring effect and the cost of maintaining the air flow rate.
When a flat sheet membrane panels with well-defined up and down flow channels were used in an aeration tank filled with mixed liquor, the membrane fouling propensity measured by TMP rising rate was much lower with fine bubbles than with coarse bubbles at identical air flow rates as shown in Fig. 1 (Sofia, 2004). The upflow velocity was found higher with fine bubbles in the particular experimental condition with unknown diffuser pore sizes, but it cannot be generalized because large bubbles are known to generate faster upflow in general. Recently in 2010, the original manufacturer of immersed flat sheet membranes (Kubota, Japan) released a new membrane frame (SP400) equipped with diffusers with smaller pores than previous (3mm vs 6-10 mm). It was claimed that the so called medium size bubbles with a larger number not only better distribute among the upflow channels between membrane panels, but also scour membranes more efficiently (Kubota Co., 2010).
Fig. 1. Effect of bubble size on membrane fouling. Air flow rates are indicated in each graph (Sofia, 2004)
On the contrary, many literatures suggest large bubbles are more effective for membrane scouring than small bubbles in flat sheet systems. Fig. 2 shows one example, where 6 g/L yeast suspension was filtered while air was supplied through the diffusers with 0.5 – 2.0 mm pores. In the range of air flow rate employed, the larger the diffuser pore size was, the lower the membrane fouling rate was under the experimental condition.
Fig. 2. Membrane fouling rate determined after two hours of yeast filtration. One A4 size module (Kubota/Yuasa, Japan) with 0.1 m2 was installed in an acrylic plastic case. The space between membrane and the case wall was kept 7 mm (Ndinisa, 2006a)
The confusing bubble size effect might be caused by the complex non-linear relationships among hydrodynamic factors. For example, if air flow rates increases in a given diffuser system, bubble sizes also increases. If air flow rates employed in experiment were high and/or the spaces between membrane surface and wall were small, even large bubbles can cover the entire membrane surface. On the other hand, if air flow rate was low, large bubbles cannot cover all membrane surfaces while small bubbles can. Under this condition, small bubbles should appear superior to large bubbles. Since the experimental condition can critically affect the conclusion when it comes to optimum bubble sizes, extreme care must be taken when generalizing the observation. Following experimental data may provide a hint of the complex nature of hydrodynamics in membrane filtration.
When flat sheet membranes (BIO-CEL®, Microdyne-Nadir) were used to filter 10 g/L yeast solutions, steady-state flux increased as air flow rate increased, but it peaked at 40 L/hr air flow rate and decreased at higher air flow rates as shown in Fig. 3 (Qaisrani, 2011). It was explained that the initial flux increase was due to the increasing bubble sizes with higher air flow rates. It was not perfectly clear about the flux decline, but bubbles were considered too large at very high air flow rates that they started to hinder the liquid to reach the membrane surface. In addition, membranes are exposed less to the wake effect of bubble due to the less number of very large bubbles.
© Seong Hoon Yoon