Partial Denitrification-Anammox Treatment of Reverse Osmosis Concentrate

INTRODUCTION AND OBJECTIVES The A.K. Warren Water Resource Facility (Warren Facility) is owned and operated by the Los Angeles County Sanitation Districts (LACSD). The facility has an average daily dry-weather flow capacity of 400 million gallons per day (mgd) and employs a high-purity oxygen activated sludge (HPOAS) process (Figure 1). Non-nitrified secondary effluent (SE) is chlorinated and discharged to the ocean. LACSD has been investigating nitrogen management options that may be implemented at the Warren Facility to address two drivers: (1) effluent reuse for the Pure Water Southern California (PWSC) program which will produce 150 mgd of purified water and (2) total inorganic nitrogen (TIN) load reduction for ocean discharge. PWSC processes that will be implemented are shown in Figure 2. Implementing nitrification-denitrification (NdN) within PWSC is projected to result in a net reduction in the TIN load discharged to the ocean (~ 49%) relative to continued operation with HPOAS. However, LACSD is proactively investigating alternative approaches to enhance load reduction beyond that achieved by implementing PWSC. Outside of conventional strategies to increase nitrogen removal, the elevated mass of nitrate in the reverse osmosis concentrate (ROC) provides a unique opportunity to reduce ocean nitrogen discharge. While uncommon, approaches considered to denitrify ROC typically incorporate biofilms and high levels of carbon dosing. This paper presents a novel approach developed by LACSD that leverages the nitrate-rich ROC and ammonia-rich HPOAS secondary effluent (SE) (Figure 2, Streams 5 and 7) utilizing a carbon efficient partial denitrification-anammox (PdNA) process. In this approach, ROC and SE would be blended at a target NH4-N/NOx-N ratio in a reactor and supplemental carbon added to drive PdNA (Figure 3). The objectives of this study were to (1) conduct proof-of-concept testing of the ROC/SE Blend PdNA process and (2) develop design criteria that can be used for planning level full-scale process sizing and cost estimation. PILOT SYSTEM AND RESULTS The ROC/SE Blend PdNA pilot system included a batch blend tank, feed pump, single-stage PdNA moving-bed biofilm reactor (MBBR), and supplemental carbon (MicroC 2000) dosing system (Figure 4, Table 1). Supplemental phosphorous (H3PO4) was added directly to the batch blend tank. Seeded MBBR media was taken from an existing mainstream PdNA pilot system that has been in operation at the Warren Facility for over a year (Sun et al., 2024). ROC was produced at the PWSC Grace F. Napolitano Innovation Center (NIC) 0.5 mgd reverse osmosis facility. However, ROC was not always available during testing due to facility shutdowns. Therefore, for part of the study, reactor feed was prepared by spiking SE with NaNO3 or by spiking NaNO3 and NaCl to target the desired NH4-INTRODUCTION AND OBJECTIVES
The A.K. Warren Water Resource Facility (Warren Facility) is owned and operated by the Los Angeles County Sanitation Districts (LACSD). The facility has an average daily dry-weather flow capacity of 400 million gallons per day (mgd) and employs a high-purity oxygen activated sludge (HPOAS) process (Figure 1). Non-nitrified secondary effluent (SE) is chlorinated and discharged to the ocean. LACSD has been investigating nitrogen management options that may be implemented at the Warren Facility to address two drivers: (1) effluent reuse for the Pure Water Southern California (PWSC) program which will produce 150 mgd of purified water and (2) total inorganic nitrogen (TIN) load reduction for ocean discharge. PWSC processes that will be implemented are shown in Figure 2.

Implementing nitrification-denitrification (NdN) within PWSC is projected to result in a net reduction in the TIN load discharged to the ocean (~ 49%) relative to continued operation with HPOAS. However, LACSD is proactively investigating alternative approaches to enhance load reduction beyond that achieved by implementing PWSC. Outside of conventional strategies to increase nitrogen removal, the elevated mass of nitrate in the reverse osmosis concentrate (ROC) provides a unique opportunity to reduce ocean nitrogen discharge. While uncommon, approaches considered to denitrify ROC typically incorporate biofilms and high levels of carbon dosing. This paper presents a novel approach developed by LACSD that leverages the nitrate-rich ROC and ammonia-rich HPOAS secondary effluent (SE) (Figure 2, Streams 5 and 7) utilizing a carbon efficient partial denitrification-anammox (PdNA) process. In this approach, ROC and SE would be blended at a target NH4-N/NOx-N ratio in a reactor and supplemental carbon added to drive PdNA (Figure 3).

The objectives of this study were to (1) conduct proof-of-concept testing of the ROC/SE Blend PdNA process and (2) develop design criteria that can be used for planning level full-scale process sizing and cost estimation.

PILOT SYSTEM AND RESULTS
The ROC/SE Blend PdNA pilot system included a batch blend tank, feed pump, single-stage PdNA moving-bed biofilm reactor (MBBR), and supplemental carbon (MicroC 2000) dosing system (Figure 4, Table 1). Supplemental phosphorous (H3PO4) was added directly to the batch blend tank. Seeded MBBR media was taken from an existing mainstream PdNA pilot system that has been in operation at the Warren Facility for over a year (Sun et al., 2024). ROC was produced at the PWSC Grace F. Napolitano Innovation Center (NIC) 0.5 mgd reverse osmosis facility. However, ROC was not always available during testing due to facility shutdowns. Therefore, for part of the study, reactor feed was prepared by spiking SE with NaNO3 or by spiking NaNO3 and NaCl to target the desired NH4-N/NOx-N ratio and TDS concentration at similar concentrations to the ROC produced at NIC. Analysis of TIN removal performance data normalized to account for carbon dose did not indicate a statistically significant difference in performance based on TDS concentration.

The pilot system was started up in April 2024 and operated for approximately 5 months. Average conditions over a four-month period of stable operation (May to September 2024) are shown in Table 2. Influent TIN and NH4-N/NOx-N ratio, nitrogen loading rates, and supplemental carbon dose are shown in Figures 5, 6, and 7, respectively. Average influent TIN, NH4-N/NOx-N ratio, and supplemental carbon dose were 74 ± 4 mg N/L, 0.45 ± 0.10 g/g, and 1.9 ± 0.45 g COD added/g TIN influent, respectively. Average influent TIN, NOx, and NH4 loading were 0.55 ± 0.05, 0.38 ± 0.05, and 0.17 ± 0.03 g N/m2-day, respectively.

Performance results are summarized in Table 3. Average TIN, TIN removed via PdNA, NOx, and NH4 removals were 57 ± 16%, 63 ± 14%, 57 ± 15%, and 61 ± 29%, respectively. Corresponding average surface area removal rates were 0.31 ± 0.09 (TIN), 0.22 ± 0.06 (NOx), and 0.10 ± 0.04 (NH4) g N/m2-day. Supplemental carbon dose varied between 0.9 — 2.7 g COD added/g TIN influent and had a significant impact on performance (Figures 8 — 10). As carbon dosing increased or decreased over this range, a corresponding increase or decrease in TIN, NOx, and NH4 removal was observed. TIN, NOx, and NH4 removals ranged between 28 to 84%, 27 to 90%, and 17 to 99%, respectively.

PdN efficiency averaged 55 ± 15% and supplemental carbon demand averaged 3.5 ±0.59 g COD added/g TIN removed. Both parameters generally improved from start-up to the end of the study (Figure 11) and over the last month of stable operation averaged 65 ± 9% and 3.1 ± 0.45 g COD added/g TIN removed. The average PdN efficiency observed in this study is within the range reported in the literature for single-stage PdNA MBBR reactors with glycerol addition: 43 ± 29% g COD/g N (Macmanus et al., 2022) and 88 ± 13% g COD/g N (Klaus et al., 2023). However, the average carbon demand observed was ~ 1 g COD/g N higher, but notably lower than the 5 — 6 g COD/g N removed for glycerol driven FdN observed by Liu et al., 2025.

Planning level cost estimates (AACE Class 5) for full-scale implementation were developed to benchmark against previously evaluated concepts (LACSD 2024). Preliminary estimates suggest that ROC PdNA may be a financially viable approach. The presentation will provide a detailed overview of the pilot system.