Intensifying BNR with low DO and DAS: The Start-up Story

ABSTRACT: This paper provides startup insights from Boulder WRRF's upgrade from MLE configuration to continuous flow densified activated sludge (DAS) with low DO without the internal mixed liquor recycle (IMLR). The DAS upgrade reduced SVI30 from 223±57 to 76±12 mL/g (66% improvement) and enabled granule formation. Profiling P release and uptake across the basins showed evidence of EBPR with potential for improved P uptake. The ongoing optimization will further fine tune the ABAC and the biological selection to improve SND and EBPR. BACKGROUND: The Boulder WRRF was recently upgraded to include hydrocyclones for physical selection and tank modifications for biological selection (without IMLR). The process changes are shown in Figure 1 and the timeline of the upgrade is shown in Table 1. Hydrocyclones were installed as a mixed liquor (ML) wasting process, with the design criteria presented in Table 2. METHODOLOGY: The densification upgrade started with installation of the hydrocyclones followed by conversion of the Z1-Z3 to anaerobic (Table 1). The hydrocyclone was operated with 20 mm and 18 mm nozzles to evaluate the impact of nozzle size on sludge settleability. To acclimate the MLSS to low DO, the aeration control in Z4 to Z8 was changed to constant low DO in May 2025 followed by ABAC in Sep 2025. Performance indicators were closely monitored using inline sensors (Figure 1) and lab analysis. Sludge samples were regularly collected to characterize the PSD, granular mass fraction, and microscopic images to study the evolution of densified sludge. RESULTS Biological selection: Figure 2 shows the selector F/M ratio during and after start-up. The F/M ratio remains above the min level required for granulation. The statistical correlation between the P uptake and the P releases was developed. P uptake was 105 percent of the P release which confirms PAO activity and EBPR occurrence, though this is below typical ratios of 1.2-1.5:1, indicating substantial opportunity for improvement. Ex-situ denitrification batch testing and P release and uptake testing are conducted to measure the specific denitrification rates and PAO activities (presented in the full paper). Settleability Improvements: The changes in sludge settleability during the densification upgrades were evaluated by comparing the SSV and SVI under different operations (Figure 3 and 4). Compared to before, SVI30 after the DAS upgrade was much lower (Figure 3). Considering the 2025 data, there was a steep decline in SVI30 from 223 ± 57 to 56 ± 10 mL/g, indicating the predominance of densified sludge with the physical selector (Figure 4). When the IMLR was turned off, the process maintained similar settleability with a higher SVI30, SSV5, and SSV30 values of 76 ± 12 mL/g, 408 mL/L, and 237 mL/L, respectively. The settleability between operations with nozzle size 20 mm and 18 mm was also compared (Figure 4). The settleability performance with both physical and biological selectors are well within the range reported for densified sludge in other studies (Regmi et al., 2025). Densification: The ML PSD was evaluated over time to track the accumulation of granular biomass (Figure 5). After DAS upgrade, there is rightward shift in particle count to include a higher percent of particles over 200 µm size indicating densification. Moreover, microscopic images of the ML and underflow show the dominant presence of dense granules (Figure 6) and the overflow wasted sludge consisiting of filamentous sludge. The full paper will delve into the change in sludge characteristics by showing the temporal changes in granule mass fraction. Outlook on continued optimization work: The plant effluent concentrations were examined to evaluate the SND and EBPR performance of the low DO process. The average effluent nitrate and TP concentrations after upgrade were 13.4 mg-N/L and 1.5 mg-P/L, respectively, exceeding the targets (10 mg/L NO3 and 1 mg/L TP). Considering P uptake and release rates across different zones, we observe an increase in PAO activity over time with different upgrades (Figure 7). While the increase in PAO inventory is encouraging, the low DO process needs fine tuning. Optimization activities will address the effluent quality issue through two lines of inquiry:
• Secondary carbon dosing: The plant has acetate dosing in the post anoxic zone. It is hypothesized that carbon dosing at the end of the process may interfere with the feast/famine cycle necessary for selecting granules capable of EBPR. Hence, the acetate dosage is minimized currently to investigate its impact on granule formation and ultimately, the effluent quality.
• Fine tuning ABAC DO set points: Comparing monthly average effluent NO3 (Figure 8), the NO3 was lower (< 10 mg-N/L) during constant low DO compared to the ABAC operation (12-14 mg-N/L). The higher NO3 values indicate that the ABAC set points are not optimal for SND. Testing of different max DO set points in a stepwise manner is underway. With the fine tuning of the DO set points, the denitrification in the process is expected to improve. Significance of study: This study delivers practical startup insights for DAS retrofits, highlighting a 6–9 month acclimation period, improved settleability, granule formation trends, and optimization of integrated low-DO/DAS systems. The paper emphasizes lessons learned, including ABAC tuning, biological selector optimization, and balancing low DO with low daily ammonia limits