Hollow fiber membrane module optimization

Specific surface area (or packing density) of membrane tank

The specific surface area can be defined either by the filtration area per membrane tank volume (m2/m3) or by the filtration area per membrane tank footprint (m2/m2). Since footprint is more of a concern in typical situations, footprint based specific surface area is commonly used to compare the packing densities of commercial membranes.

Large specific surface area not only allows small tank size/footprint, but also allows power savings for scouring air and cleaning chemical savings, etc. (Chang, 2011). Hollow fiber membranes are more compact than flat sheet membranes in general. The average specific surface area based on footprint are reported at 292 m2/m2 and 118 m2/m2 for commercial hollow fiber and flat sheet membranes, respectively (Santos, 2011). These packing density allows HRTs somewhat less than 1 hr for hollow fiber and around 2.0 hr for flat sheet.

According to a proposal submitted to the City of Visilia in May 2012, GE’s newest hollow fiber technology offers 263 m2/m2(or ft2/ft2), where up to 480 ZW500d modules with 370 ft2 (or 34.4m2) are packed in 10 cassettes that are placed in a 75 feet long, 9 feet wide, and 13 feet high tank. For flat sheet membrane, according to a blueprint available here, Kutota’s technology offers around 100 m2/m2, where 4,800 flat sheet membranes with 1.25 m2 are packed in 12 cassettes (aka submerged membrane unit, SMU) that are placed in a 14 m long and 4.5 m wide tank. Other blueprint available here also shows similar design factors for flat sheet membrane.

Many flat sheet immersed membrane modules are stacked upward in order to improve specific surface area per footprint. As a result, flat sheet cassettes tend to be taller than hollow fiber cassettes, e.g. Kubota’s SP400 vs GE’s ZW500d. However, the maximum stacking height exists due to the inefficient membrane scouring caused by the bubble coalescence, upflow channelling, and the maintenance challenges.

Effect of membrane packing density on module performance

The productivity of hollow fiber membranes is related to both membrane packing density as well as flux. A high membrane packing density would increase productivity as long as flux is not detrimentally affected (Yeo, 2006). However, it has been known that excess packing density causes high solids flux toward the membrane bundle, which causes high solids level near the membrane surface. Flux can decrease as a result of the particle accumulation inside the fiber bundle. Though the particle accumulation can be controlled to some extent by increasing scouring air flow, the increased energy cost may not justify the high packing density. Therefore, the optimum membrane packing density must be determined considering the target MLSS, target flux, fiber movement as a function of scouring air flow, etc.

While particle-free liquid is permeating through the fibers, particles are accumulating inside the hollow fiber bundle. In normal condition, particle concentration inside the bundle is balanced at slightly higher level than that in bulk as a result of the enhanced mass exchange by the transversal flow through the fiber bundle. However, the transversal flow velocity reduces if packing density increases. The reduced transversal flow causes high MLSS in the fiber bundle, which in turn causes high medium viscosity. Eventually the slow transversal flow and the high MLSS can lead to detrimental fiber clogging. Therefore, the high fiber density establishes the environment with high membrane clogging potential in general.

Kiat et al. (1992) assessed the effect of the packing density on flux. Their experimental results showed that serious inter-fiber clogging occurred when the packing density of the module exceeded a critical value (about 20 threads cm2) under the condition employed in the experiment. However, the critical packing density should vary depending on the type of feed water and the dimension and the material of membrane fibers. Since fiber clogging is a consequence of the imbalance between the particle accumulation in membrane bundle and the lateral flow velocity generated by the turbulence generated by rising bubbles, the optimum fiber packing density must be dependent on the hydrodynamic conditions around the fiber.

Fig. 1 shows a most commonly used hollow fiber membrane cassette in MBR, where hollow fibers are potted in relatively narrow bottom header to enhance mass transfer in and out of the bundles. In addition equal or bigger spaces than the bundle depth exist in between membrane bundles. Due to the slack (or extra length) given to the hollow fiber bundles, membrane fibers are more equally distributed in the middle of the height. Overall the fiber density per footprint is no more than 15%.

Packin100Fig. 1. Hollow fiber modules installed in a cassette (GE).


© Seong Hoon Yoon