Membrane fouling roadmap – Sudden TMP rise

The sudden TMP rise (or jump) shown in Fig. 1 is commonly observed in MBR and surface water filtration, where flux is maintained largely at a constant level while TMP is allowed to go up to compensate the permeability loss of membrane. This phenomenon can be best explained by cake layer compaction (Chang, 2006; Park, 2006; Fane, 2007). In this theory, once pressure loss through the cake layer reaches a critical level, cake layer compaction starts from the bottom of the cake layer due to the cumulative nature of the downward force in the cake layer as explained here. As cake layer compaction proceeds, TMP must rise to compensate the permeability loss under constant flux mode, which in turn accelerates cake layer compaction. As a result, TMP rises exponentially as shown in Fig. 1.

There are other theories explaining the sudden TMP rise.

  1. Pore loss (or blocking) model

As pores are blocked by foulants as a consequence of membrane fouling, more water has to pass through other remaining pores under a constant flux mode. In the beginning, the lost water permeation can be easily compensated by the large number of intact pores by a slight flow rate increase. However, toward to end of the filtration cycle, not enough number of intact pores are left over. As a consequence, loss of intact pores impact increasing more on the blocking (or fouling) of remaining intact pores by particles. Once the flux to the remaining pore reaches a critical level, pore loss accelerates and TMP rises quickly (Ognier, 2004; Ye, 2006).

  1. Area loss model

This model is based on similar idea as the pore loss model above. It suggests membrane fouling does not proceed evenly across the surface area, which causes certain areas fouled quicker than other areas. Once one area is fouled first, the flux in relatively clean area must go up in order to maintain a constant apparent flux. Eventually membrane fouling spreads across the membrane surface in relatively short period of time (Cho, 2002). This area loss model is particularly plausible in large-scale MBR plants, where diffuser fouling can cause a severe fouling in the membranes above the diffusers. The lost permeate from the fouled membranes must be compensated by other relatively clean membranes, which can trigger accelerated TMP rise across the board.

  1. Percolation model

Macromolecules passing the cake layer can deposit inside the cake layer and reduce cake porosity. Once cake porosity reaches a threshold level, permeability of the cake reduces sharply decreases and TMP must rise to keep a constant flux (Hermanowicz, 2004).

  1. Osmotic pressure model

In reverse osmosis, once ions pass cake layer and rejected by membrane, back diffusion to the bulk is restricted due to the hindrance of cake layer. As a consequence, ions tend to accumulate near the membrane surface more than the level predicted by film theory (Hoek and Elimelech, 2003). Likewise, the cake layer formed by EPS and SMP on MF/UF membrane can restrict the back diffusion of charged molecules in the cake layer (Fane, 2009). However, it is yet to be found out how much osmotic pressure can be generated by the relatively large molecules rejected by the dynamic membrane.

  1. Quorum induced biofouling and cake compaction

Microorganisms communicate each other using small molecules called “quorum” and change metabolic activity/state as a whole group depending on environment. There are many different kind of quorums, but N-acyl homoserin lactone (AHL) has been identified as a main quorum related with cake layer compaction (Lee, 2009). There is no firm consensus on this theory, but it is interesting to see that AHL content in the cake layer of hollow fiber membranes is strongly correlated with TMP as shown in Fig. 2. Since the amount of cake layer does not increase proportionally with TMP, high ALH content directly indicates high specific AHL content as shown in Fig. 3.

In spite of the observations above, it is not clear whether cake layer compaction caused high AHL level or high AHL level caused cake layer compaction. However, same research group showed that membrane permeability loss could be successfully mitigated by so called quorum quenching, where AHL is hydrolyzed by acylase (Kim, 2010). The result obtained with a pure culture (P. aeruginosa) suggested AHL can be a cause of not only cake layer compaction, but also cake layer formation itself to some extent. In this study, the NF membrane with a immobilized acylase I suffered much less permeability loss than control membrane. With immobilized acylase I, the porosity of cake layer measured by CLSM showed was substantially higher (Fig. 4) and the relative flux was also maintained higher (Fig. 5) comparing to those of raw membrane.


Fig. 1. Typical TMP rising pattern in submerged membrane process. Flux = 15 LMH, Yuasa’s flat sheet immersed membrane,  MLSS = 12 g/L, 20 oC (unpublished data, Yoon).


Fig. 2. Correlation between biofouling and the content of N-acyl homoserin lactone (AHL) in cake layer formed on hollow fiber membranes in MBR, where the depth of green color of the indicating agar is proportional to the AHL content in the cake layer on used hollow fiber membranes (Lee, 2009).


Fig. 3. Specific AHL level in cake layer as a function of the amount of biocake (Lee, 2009).


Fig. 4. Spatial distribution of areal porosity of P. aeruginosa biofilm cultivated on the raw NF (black circle) and Acy-NF membranes (red triangle) in the flow cell unit: (a) based on cells and (b) based on polysaccharides (Kim, 2010).


Fig. 5. Comparison of flux profiles between the two NF systems equipped with membranes with and without Acylase enzyme immobilization (Kim, 2010).


© Seong Hoon Yoon