Performance Evaluation of an Online Membrane Integrity Monitor for Flat Plate MBRs

Incorporating membrane bioreactors (MBR) into potable reuse treatment trains is gaining interest due to increasing concerns over water scarcity and the need for high quality effluent for downstream treatment technologies. However, membranes are vulnerable to breaches which necessitates real-time monitoring of membrane integrity. Turbidity monitoring is applied ubiquitously in the field, but it is difficult to correlate with microbial reduction performance. To this end, a novel online integrity monitor was developed and evaluated at a 30,000 gal/d MBR pilot in Canton, OH. This monitor detects changes in permeate quality by quantifying the fouling of a dead-end membrane external to the MBR treatment process via flow and pressure measurements. Figure 1 details the configuration of the integrity monitor. The performance of the integrity monitor was validated by intentionally creating membrane breaches. Slits were cut into the membranes of one module prior to installation and were protected with tape and tied to strings that could be removed by operators for control of breach timing. The intensity of damage was controlled by varying the number of slits in three phases separated by 24 hours; three slits (Phase 1), followed by 28 slits (Phase 2) and finally 63 slits (Phase 3), totaling 94 slits (Table 1.). Membrane integrity was monitored using two online instruments: a commercially available laser-based turbidimeter and the novel membrane integrity monitor. The integrity monitor provided two parameters used for comparison with turbidity including the resistance (transmembrane pressure/flux) and the graphical first derivative of the resistance. Log reduction values (LRV) of male specific and somatic coliphages, E. Coli and total coliforms were assessed before and after breaches for comparison with the real-time data. Baseline MBR effluent turbidity was <0.04 NTU; after Phase 1 damage was uncovered, turbidity briefly spiked above 1.0 NTU, but recovered to pre-damaged conditions. Initial results (Figure 2) suggest that membrane breaches can be reliably detected by the prototype integrity monitor. Breaches were detected when the derivative of the resistance (αR) was > 1.05; turbidity and microbial results corroborated the response of the integrity monitor. The resistance increased positively over the three phases from 0.5 to > 8.0 psi/gal/ft2/d while αR detected the second phase breach (28 slits) with a marked response (Figure 2). It is possible that the three slits of Phase 1 did not produce a response because it was simply too small of a breach; it is possible that the mixed liquor may have provided a 'band-aid' effect in where the slits were covered by mixed liquor during filtration, disallowing particle passage into the permeate. Work is ongoing to determine the lower limits of sensitivity of the monitor by varying parameters and configuration. LRVs of male specific coliphages fell from approximately 3- to < 2-log but recovered after approximately one hour of operation. Total coliform and E. Coli LRVs reduced from 7.2 and 6.7-log, to 4.2 and 3.8-log, respectively; the LRVs remained stable after this initial drop. Microbial response to the breaches can be seen in Figure 3. This presentation and paper show progress towards developing a sensitive, integrity monitor that will reliably detect membrane breaches in real time. Work is ongoing to improve the sensitivity of the device, identify patterns in the data it produces, and correlate its results to microbial reduction performance.