Evaluation of cake filtration biological reactors (CFBR) vs. membrane biological reactors (MBR) in a pilot scale plant

Abstract

In this study, cake filtration biological reactors (CFBR) and membrane biological reactors (MBR) were operated in the same environment. The initial flux of CFBR was 1,517.5L/h/m2 and the start-up cake resistance was around 6kPa. The flux dropped sharply during the first run, and at the end of first run flux was 110L/h/m2. During the second run, the flux declined further from 60L/h/m2 to 48L/h/m2 and during the last 4,000hours period flux was considered to be stable. However, the pressure difference (∆p) increased steadily for about 2,300hours until it reached a stable ∆p value at the termination of the run. On the other hand, the design operation flux of the membranes was 17.5L/h/m2. The system was operated around design flux values; however, more stable operation was achieved when the flux was around 9L/h/m2. Both the cloth filter and the membrane produced effluents with suspended solid concentrations of less than 10mg/L, but the effluent quality of the cloth filter was inferior to the membrane with respect to coliform removal. Activated sludge cake, which formed on cloth media during filtration, was evaluated according to conventional cake filtration theory using the plots of V vs. t and t/V vs. V. The plot of V vs. time t for filtration of different MLSS concentrations in laboratory cell was best fitted to second order polynomial regression (R2>0.98); a linear relationship between t/V against V was not observed. Separately, an evaluation of V vs. t showed that CFBR was best fitted to second order polynomial regression (R2=0.995 for first run and R2=0.989 for second run); however, a linear relationship between the reciprocal of filtration rate (t/V) and the cumulative volume was observed individually. A transition between two linear plots was considered to be a change in the filtration characteristics. Standard blocking model plots of CFBR were fitted better (R2>0.95) than complete blocking and intermediate blocking models with the measured data for the initial period of filtration; the latter period of filtration was best fitted to the cake filtration model (R2>0.99).