MABR Breathing Insights: Monitoring Biofilm Health in Full-scale System Based on Exhaust Gas Measurements

Introduction:
With the current needs for balancing process intensification and operational cost, Membrane aerated biofilm reactors (MABR) stand out as an advanced intensification technology. Using gas-permeable membranes, MABR is bubbleless aeration technology where oxygen is effectively delivered to the biofilm grown on the biofilm surface (Houweling & Daigger 2019). Uniquely, MABR's gas-substrate counter diffusional pattern provides the opportunity to obtain 'Breathing Insights' by assessing the oxygen transferred to the biofilm which can be utilized to monitor, diagnose, and control biofilm and process performance. To this end, studies reported that the relationship between Oxygen purity in the exhaust gas and ammonia concentrations can be used to measure the response from the MABR to ammonia loads which indirectly indicates biofilm health or nitrification capacity (Houweling & Daigger 2019; Uri-Carreno et al. 2021). In accordance, a soft sensor named 'MABR fingerprint' was proposed based on the response of the MABR to the diurnal changes in the bulk liquid ammonia concentration (Yang et al., 2022; Cao et al., 2023). In such studies, the fingerprint concept was assessed based on the outcomes from steady state and dynamic simulations. However, validation of such concepts has not been performed in full-scale systems. In addition to the MABR fingerprint, based on exhaust oxygen purity, OTE and oxygen transfer rates (OTR) can be determined (Houweling & Daigger 2019). The obtained OTR can be translated into maximum nitrification rates (Max_NR) of the MABR. Hence, monitoring and comparing the actual daily nitrification rates (NR) with the Max_NR can, also, be used as a simple indication of biofilm nitrification activity. Therefore, in this study, we leverage the data from one of the largest ZeeLung™ MABR installations in North America, to evaluate the feasibility of using MABR fingerprint and NR/Max_NR ratio to monitor and diagnose MABR performance.

Operational information:
The data used in this study is from Hespeler WRRF, Cambridge, Ontario, Canada. The data covers a period of 5 months grouped into 5 phases as described in Table 1. Significantly, biofilm thickness was measured three times during the 5 months which had a clear gradient as biofilm thickness decreases over time. Detailed information on biofilm thickness are provided in (Lakshminarasimman and Parker, 2025).

MABR fingerprint plot:
The fingerprint plot is obtained by plotting the daily data of the Exhaust oxygen Purity (%) vs the bulk/effluent ammonia concentrations. In prior studies where data were obtained from model simulations, high correlation (above 0.9) was obtained between both variables (Cao et al., 2023). Accordingly, the fingerprint slope, referred hereafter as the daily slope, was obtained. Such fingerprints plots were not consistently observed with the actual daily, where correlations ranged between 0.5 and 0.9. Dirunal dynamics yielded two major shapes, presented in Figure 1. The Fingerprint was grouped into three major periods. The first is when the influent ammonia is increasing (the loading period), and two periods when bulk ammonia is decreasing (the recovery period). In addition to the daily slope, each period's slope was determined separately.

Monitoring MABR:
By examining the generated plots over different periods, the shape with exponential decay was observed mainly during the first two phases where relatively low ammonia loading rates were applied in comparison with phases III and IV. This can be referred to mass transfer limitations due to the very thick biofilm thickness (850-1000 micron), such effect was not mitigated when air flow rate increased to 6 Qair/m2/d from 4.5. Such a pattern started to disappear and an inclined line was observed when biofilm thickness started to be reduced in the following phase. This can be leveraged as a useful tool to control scouring and air flow rates to either reduce biofilm or overcome mass transfer limitation by higher air flow rates.

In Figure 2, the obtained slopes from different phases are presented. Clearly lower slopes were obtained at the phases closer to the time biofilm thickness decreased to 450 micron (Phase IV and V). It can be seen that such effect was captured clearly in the loading slope and to much lesser extent in the daily slope. A steeper slope indicates better response from the biofilm to ammonia load, especially when fingerprint plot are closer to those in Figure 1a.

In Figure 3, NR/Max_NR for different phases is plotted. It might not be as representative to biofilm thickness as the finger print plot, but it provides useful information on oxygen utilization by nitrifiers in the system. Phase I and V had low NR/Max_NR, but in the case of Phase I, this was due to the very thick biofilm thickness and low ammonia loads which probably led to higher growth of heterotrophs and other oxygen utilizing organisms. This is in agreement with microbial analysis performed, results not published. On the other hand, low NR/Max_NR at Phase V is mainly due to low ammonia loading rate which is confirmed from the fingerprint plot that shows relatively steep slopes.

Conclusion & significance:
This is the first study to examine, in full-scale with real-life data, off-gas based metrics to monitor MABR biofilm health and performance. This provides simple and easy to understand tools not only to monitor the performance but also to inform decisions.