0.6 MGD Validation of a Unique Tertiary MBR Process to Provide NdN for Pure Water

INTRODUCTION
The Los Angeles County Sanitation Districts (LACSD) and the Metropolitan Water District of Southern California (MWD) are jointly developing the Pure Water Southern California program, which would supply up to 150 million gallons per day (mgd) of potable reuse water to the region from LACSD's A.K Warren Water Resource Facility (Warren Facility). The Warren Facility is a 400-mgd capacity non-nitrifying High Purity Oxygen Activated Sludge Plant currently being evaluated for a 'Flex membrane bioreactor (MBR)' concept that would employ a modular design enabling different reactor modes as needed by the program (Figure 1; Fitzgerald et al., 2024).

One potential mode is nitrification-denitrification (NdN) tertiary MBR (tMBR) using glycerol as the carbon source. This mode would offer additional treatment barriers between wastewater and the product water but at higher operating cost. Initial design utilizing a post-anoxic configuration showed nitrite (NO2) accumulation when tested at demonstration scale (Lai-Bluml and Trussell, 2020). Later designs utilizing a pre-anoxic configuration showed no such issue at bench and pilot scale (Liu et al., 2022; Deco et al., 2023). The current work validates these findings at demonstration scale (~0.6 mgd).

METHODOLOGY
Testing was conducted at the Napolitano Center's Demonstration Plant. The process' flow diagram is illustrated in Figure 2 and was seeded with mixed liquor from a local MBR. Warren Facility secondary effluent was utilized as the feed with characteristics summarized in Table 1. The glycerol source was MicroC-2000 (EOSi, Pocasset, MA). Additional design and operational parameters are summarized in Table 2.

The study included two major phases (Table 3): Phase 1 involved constant flow operation; Phase 2 involved diurnal flow operation, corresponding to program buildout. Routine monitoring results are summarized in Table 4. 16S rRNA metagenomic analysis was performed by DNASense (Aalborg, Denmark).

RESULTS
Observations Fitting Expectations
Figure 3 summarizes MBR filtrate nutrient levels from the study. During constant flow operation (Phase 1), all metrics met the targets: ammonium (NH4) and NO2 were consistently below 1 mgN/L; nitrate (NO3) was at or below 19 mgN/L; and soluble orthophosphate (SOP) was below 1 mgP/L. During initial diurnal flow operation (Phase 2a), process instability was encountered (discussion below). After re-stabilization, diurnal flow operation was revisited with enhanced dissolved oxygen (DO) monitoring (Phase 2c) and again met the targets.

Denitrification stability was judged by anoxic NOx (NO2+NO3) residual, which can correlate with NO2 accumulation (Liu et al., 2022). Anoxic NOx remained stable and low throughout the study (Figure 4), except: (a) during the process instability event (Phase 2a); and (b) when the system was stressed for lower NO3 target (Phase 2c). Carbon demand in chemical oxygen demand (COD) added to nitrogen removed (C/N) was in the expected range of 5~6 g-COD/g-N (Figure 5).

Unexpected Observations
Biological Instability Event
Diurnal testing (Phase 2a) showed periodic NH4 bleed through (Figure 3). This issue was attributed to a faulty DO probe and under-aeration. Airflow was increased in June which addressed NH4 bleed through but led to NO2 accumulation (Figure 6) and broad process instability, including drops in mixed liquor total and volatile suspended solids (MLSS and MLVSS) (Figures 7 and 8) and increases in filtrate SOP (Figure 9), COD (Figure 10), and time-to-filter (TTF; Figure 11). Influent inhibition was ruled out as a cause as the symptoms did not appear in other nitrifying pilots operated in parallel.

The instability event significantly reduced NO2 oxidation, while minimally affected NH4 oxidation (Figure 12). 16S rRNA metagenomic analysis showed a dramatic shift in nitrite-oxidizing bacteria (NOB) population from comammox-dominated to canonical NOB (Figure 13).

Diurnal testing was revisited with enhanced DO monitoring. During this period, the instabilities did not return, suggesting under-aeration as a potential cause for the instability. Additional microbial analyses are underway.

Biomass Yield The modeled biomass yield was ~0.3 lbs-VSS/lb-CODadded, but observations showed ~0.4 lbs-VSS/lb-CODadded (Figure 14). Two hypotheses were proposed: enrichment of specialist heterotrophs and internally stored carbon (ISC), as supported by the following:
- Ordinary heterotrophic organisms (OHOs) typically have a decay rate of 0.67 d-1, while specialist organisms, like methanol utilizers, have lower decay rates near 0.05 d-1. Analysis indicated that a specialist population with a decay rate of ~0.1 d-1 could explain the higher yield observed (Figure 15).
- Microscopy with Neisser staining suggests enrichment of organisms with storage granules (Figure 16).
- Increased airflow during process instability led to a loss of MLSS (Figure 7) and a reduced MLVSS/MLSS ratio (Figure 8), possibly due to rapid ISC oxidation.

BENEFITS AND SIGNIFICANCE
This work demonstrated that pre-anoxic NdN TMBR successfully addresses NO2 accumulation previously observed and met all key performance targets. The instability event encountered may have broader program and industry implications. Most notably, comammox-dominated nitrifying and ISC-driven denitrifying systems may be created with under-aeration and destabilized with a sharp increase in aeration.