Longevity of membrane in MBR – Third party’s estimation

For membrane manufacturers, it is easy to draw a line between live and dead membranes since it can be judged by membrane replacement as discussed here. However, there is no clear cut in membrane life span for membrane users. From a practical point of view, membrane life ends when it cannot treat peak flow due to low permeability. But, membranes often extend their life with a help of extra recovery cleanings with chlorine and/or organic acids. More often than not, oversized membrane capacity also helps extend membrane life by providing surface areas during peak flow. If lucky, nearby wastewater plant may take excess wastewater that cannot be treated by MBR, too.  

Fig. 1 shows the breakdown of the reasons of membrane replacement.

Membra3aaFig. 1. The causes of membrane replacement (Ayala, 2011)

¬†¬† Following are four different criteria that provide hints on membrane life expectancy. Ayala et al (2011)’s study is based on the survey and the experiments performed with flat sheet Kutota membranes.¬† Fenu et al (2012)’s study is based on a survey performed in a MBR plant with GE ZW500c membranes in Schilde, Belgium, where excess wastewater is directed to a nearby conventional activated sludge plant.

1.  Based on permeate flow rate

¬†¬† If membrane’s life span is judged based on the capability of producing design flow rate, the membranes could be called “failed” at around 4.5 years (or 1,650 days), where the MBR failed to produce the design flow of 220 m3/d in winter. Excess wastewater was directed to a nearby plant during the time. However, the MBR produced near design flow in the next winter as a consequence of more frequent recovery cleaning. In this case, if the membrane failure was judged based on design flux, a pre-mature membrane replacement could occur.

   One more complexity of judging the membrane failure is that the low flow event could be caused by simply a temporary deterioration of mixed liquor quality. Therefore, even if membranes fail to produce design flow, it is hard to conclude that membranes need replacement.

Membra23Fig. 2. Flow, fouling rate, and temperature in a MBR plant in Schilde, Belgium (Fenu, 2012)

2.  Based on permeability

     Permeability was monitored since the MBR was commissioned in the same place mentioned in above paragraph. As shown in Fig. 3, membrane permeability fluctuates while declining overall. The high points are obtained during summer and/or after recovery chemical cleaning while the low points are mainly obtained during winter and/or in the end of a filtration cycle. One thing noticeable is that the high points and low points are converging each other over time. This means that the efficacy of membrane cleaning declines over time due to the accumulation of irreversible membrane fouling or permanent membrane damage. It can be assumed that membrane cleaning becomes very ineffective when membrane needs replacement. It is not straightforward how to draw the curves for high and low points, but the two curves in Fig. 3 appears converge at around 4,000 days or 11 years.

Membra24Fig. 3. Determination of the timing of membrane replacement (Fenu, 2012).

3.  Based on chlorine exposure

   The recommended maximum chlorine exposure of ZW500 membranes is 500,000 ppm.hr. In the Schilde MBR, maintenance cleaning was performed weekly while recovery cleaning was performed whenever TMP reaches 0.4-0.5 bar. Chlorine concentration in cleaning solution was at 600 mg/L as NaOCl.

   Over the first 10 years, cumulative chlorine exposure (green bars) is exponentially increasing due to the increasing membrane cleaning frequency. If the curve is extrapolated, cumulative chlorine exposure reaches the maximum limit of 500,000 ppm.hr in the 15th year of the commissioning.

Membra25Fig. 4. Yearly and cumulative chlorine exposure (Fenu, 2012).

4. Based on pure water flux

   The clean water fluxes (CWF) of flat sheet Kubota membranes used in many different MBR plants for various period time were measured. After cleaning the used membranes using NaOCl, CWT was measured at a static pressure of 0.5 m H2O at 20oC. The measured CWF was normalized against the CWT of virgin membrane.

   As shown in Fig. 5, the normalized CWF clearly declines as membrane age increases. Interestingly, the CWF after membrane cleaning was always higher than that of virgin membrane, if the membranes were used less than 6 years (or 72 months). Normalized CWF was less than 90% only when the membrane used for 8 years. It was postulated that the gradual CWF loss was caused by pore plugging by inorganic precipitates that are hardly removable by NaOCl or organic acids. (Note: New membranes often come with loosely bound residual chemicals from synthesis that can be easily removed by cleaning chemicals, which can lead to higher CWF after a few first cleaning cycles)

   The normalized flux lower than 100% does not necessarily mean membrane must be replaced. Therefore, this result only suggest that flat sheet membranes can be used at least 6-8 years from the perspective of CWF.

Membra4Fig. 5. Membrane age vs. flux of the chemically cleaned membrane (Ayala, 2011)

5.  Conclusion

  • There is no decisive criterion that triggers end of membrane life, but the capability of obtaining design flux within the maximum trans-membrane pressure (TMP) with a help of chemical cleaning at reasonable frequency can be a criterion in practice.
  • In Schilde MBR, membrane longevity estimation of the least performing train out of four trains was 8.1 – 10.0 years using the above criteria.
  • The estimation based on chlorine exposure appears overly optimistic (15 years).
  • Overall, above estimations are not totally different from the manufacturer’s estimation.


© Seong Hoon Yoon