According to the data available from the public domain, small bubbles are more effective for short hollow fiber membranes (110-400 mm) and tightly held long hollow fiber membranes (820 mm)than large bubbles at a same air flow rate. However, large bubbles appear more effective for long hollow fiber membranes held loosely, e.g. ZW500dTM, SADFTM, etc. The exact reason is not clear at this point, but it might be related with the fiber movement in shear field that is a crucial part of the anti-fouling mechanism in addition to direct scouring by bubbles.
When a single hollow fiber with 1 mm OD and 400 mm length was installed tightly in a square channel (20 mm x 20 mm) and superficial liquid velocity was controlled by a circulation pump, slower membrane fouling was observed with small bubbles that with large bubbles at a same air flow rate. As shown in Fig. 1a, membrane fouling rate was the lowest with small bubbles at a low superficial liquid velocity (16 mm/s). When the liquid velocity was raised to 160 mm/s, however, the bubble sizes effect became less apparent as can be seen in Fig. 1b.
Similar bubble size effects were observed in many other studies. When shear stress on membrane surface was measured using multiple 120 mm hollow fiber membranes, the high number of weak shear events generated by small bubbles resulted in more favorable hydrodynamic conditions for fouling reduction and flux enhancement (Chan, 2007). Fine bubbles were more effective in membrane fouling mitigation than coarse bubbles in other experiment performed with five 820 mm long fibers held tightly (Martinelli, 2010). Fine bubbles were more effective in ultrafiltration of river water using 110 mm long membrane fibers (Tian, 2010).
Fig. 1. Membrane fouling rate measured by TMP increasing rate at different linear liquid velocity: a) 16 mm/s, b) 160 mm/s. A single hollow fiber membrane with 1 mm OD and 400 mm length was used to filter 2 g/L bentonite (Yeo, 2007).
In contrast, when hollow fibers with 0.65 mm OD and 500 mm length were held with 1% looseness (relative excess fiber length to the distance between two headers), no significant differences were observed between 0.5 mm and 1.0 mm bubbles with regard to membrane fouling (Wicaksana, 2006). In commercial membranes with ~2,000 mm length held with 2.5% looseness, coarse bubbles were superior to fine bubbles (Hong, 2010).
It is not clear why the bubble size effect turned out oppositely depending on literature, but it is likely because direct contacts of bubbles with membrane fibers are becoming less dominant antifouling mechanism as fiber length grows especially when fibers are held loosely. In fact, bubbles tend not to rise along the fiber due to the higher resistance imposed by the fiber itself, but rise through the spaces between fiber bundles. In this case, random lateral fiber movement caused by┬áthe rising bubbles becomes the major antifouling mechanism. Since large bubbles displace more water volume, which in turn induces more vigorous water stream meandering, they are more effective in inducing the random lateral hollow fiber movement. Therefore, all the known commercial hollow fibers are using coarse bubble diffusers with 6-10 mm pore size.
In summary, small bubbles tend to be more efficient than large bubbles for flat sheet membranes, tightly held hollow fiber membranes, and short hollow fiber membranes held loosely. For these types of membranes, the direct contact of bubbles with membrane surface and the upflow velocity generated by rising bubbles are the major membrane aniti-fouling mechanism. On the other hand, large bubbles are more effective for commercial scale hollow fiber membrane modules, where random fiber movements induced by rising bubbles are the major antifouling mechanism. The optimum bubble size must be experimentally determined for each specific membrane module configuration due to the insufficient understandings about the bubble dynamics and the hydrodynamic behaviour of mixed liquor.
┬ę Seong Hoon Yoon