Lessons from the history

Philosopher Georg Hegel  quotes “We learn from history that we do not learn from history”. It seems largely true in social science, but fortunately it is largely untrue when it comes to engineering. We learn from the history.


The current MBR technology with immersed (or submerged) membranes is an outcome of numerous trial-and-errors that had undergone during the last two decades. Many misunderstandings, under- and overestimations, and misguidance have been corrected. Following are some of the most prominent ones.

  •  As membranes were considered compatible with non-biological debris, the necessities of fine screens to remove debris were often underestimated in early days. The clogging/ragging of membranes by fibrous materials, especially for hollow fiber modules, is detrimental for the entire MBR process. The debris stuck in hollow fiber bundles can only be hand picked while mechanical surface scouring after taking all membrane panels from the frame is often an only method in flat sheet membranes. Some relevant pictures are provided here. This issue has been largely solved when mechanical screens with 2-3 mm pore size or less was introduced. However, this problem is still commonly experienced due to the improper screens or glitches in installation (Stefanski, 2011).
  • Focusing on reducing plant footprint, aeration basins were often designed at too high MLSS such as 20 g/L or even 30 g/L. The high MLSS not only can hamper membrane scouring by slowing down upflow, but also can increase biopolymer concentrations by hampering oxygen transfer. In addition, the slow mass transfer in and out of the membrane bundle can cause hollow fiber clogging. As a consequence, excessive membrane fouling was commonly experienced in the plant designed for high MLSS. The design MLSS has decreased significantly since late 1990’s. Now the optimum MLSS in aeration basin is considered 8-12 g/L while some MBR are designed at even lower MLSS at 6 g/L.
  • MBR was thought to be compatible with high F/M ratio. As membrane rejects solids 100%, solids settling in clarifier became no longer a concern in MBR. As a result, the F/M ratio, which was controlled low in conventional activate sludge (CAS) for a good biosolids settling, was not considered a significant factor in MBR. This misperception was combined with the desire of saving footprint, which resulted in overly compact aeration basins. The resulting high F/M ratio not only caused low oxygen transfer efficiency (OTE), but also increased membrane fouling rates by increasing biopolymer concentrations. Nowadays F/M ratio of MBR is considered 1/3 to 1/2 of the CAS as discussed here.
  • MBR was considered tolerable to organic loading shock. In CAS, organic loading rate must be maintained stably as much as possible to obtain good sludge settling. But it was thought to be an obsolete concept for MBR since no clarifier exists. As a result, it was thought reducing or eliminating holding (or equalization) tank was plausible. It is somewhat arguable, but widely varying F/M expedites membrane fouling in many situations. If DO is not maintained high enough during the high organic loading, membrane fouling becomes even more significant.
  • Biosolids production (or sludge yield) of MBR had often been claimed much lower than that in conventional activated sludge (CAS) process. This claim was perhaps based on some lab- or pilot-scale experiment performed in the dawn of MBR technology in early 1990’s or earlier, where unrealistically high SRT such as 50-100days was employed. In addition, the commercially motivated desire to highlight the advantage of MBR technology perhaps propelled this half truth. However, field engineers soon realized there was little difference in apparent biosolids production from MBR and CAS by early 2000’s. Although MBR produces slightly less biosolids due to its longer SRT (12-30days for MBR vs 5-10days for CAS), it does not lose biosolids through the effluent. As a result, the apparent excess biosolids productions are not much different and often not distinguishable. Fig. 1 shows the simulation data based on ASM#1 assuming varying degrees of suspended solids loss from CAS, where apparent Yobs are about same in the common SRT ranges of CAS and MBR.

Fig. 1. Comparison of observed sludge yield in CAS and MBR (unpublished data, Yoon 2003)


© Seong Hoon Yoon