ABSTRACT
The algal–bacterial shortcut nitrogen removal (ABSNR) process can be used to treat high ammonia strength wastewaters without external aeration. However, prior algal–bacterial SNR studies have been conducted under fixed light/dark periods that were not representative of natural light conditions. In this study, laboratory-scale photo-sequencing batch reactors (PSBRs) were used to treat anaerobic digester sidestream under varying light intensities that mimicked summer and winter conditions in Tampa, FL, USA. A dynamic mathematical model was developed for the ABSNR process, which was calibrated and validated using data sets from the laboratory PSBRs. The model elucidated the dynamics of algal and bacterial biomass growth under natural illumination conditions as well as transformation processes for nitrogen species, oxygen, organic and inorganic carbon. A full-scale PSBR with a 1.2 m depth, a 6-day hydraulic retention time (HRT) and a 10-day solids retention time (SRT) was simulated for treatment of anaerobic digester sidestream. The full-scale PSBR could achieve >90% ammonia removal, significantly reducing the nitrogen load to the mainstream wastewater treatment plant (WWTP). The dynamic simulation showed that ABSNR process can help wastewater treatment facilities meet stringent nitrogen removal standards with low energy inputs.
HIGHLIGHTS
Sidestream wastewater treatment by algal–bacterial shortcut nitrogen removal.
Varying illumination simulated summer/winter outdoor conditions.
Model developed for dynamic PSBR simulation under varying illumination.
Long-term simulation employed for TN removal in a full-scale WWTP.
INTRODUCTION
Anaerobic digestion (AD) sidestreams contain high ammonium () concentrations and have low organic carbon to nitrogen (C/N) ratios. These wastewaters are often recycled back to mainstream wastewater treatment processes, making it difficult for facilities to meet stringent nutrient discharge limits. AD sidestreams are costly to treat using conventional biological nitrogen removal (BNR) processes due to their high oxygen (O2) demands for nitrification and lack of sufficient organic carbon for denitrification. The shortcut nitrogen removal (SNR) process has been developed to reduce O2 and organic carbon requirements by suppressing the growth of nitrite-oxidizing bacteria (NOB) that transform nitrite (
) to nitrate (
). NOB suppression is accomplished by maintaining high free ammonia (FA), high free nitrous acid (FNA) and/or low dissolved oxygen (DO) concentrations in the bioreactor (Anthonisen et al. 1976; Wang et al. 2015). The use of SNR results in a 25% reduction in aeration requirements for
to
oxidation and a 40% reduction in organic electron donor requirements for
to N2 reduction and reduced sludge production compared with conventional BNR (Peng & Zhu 2006).
Integrating an algal–bacterial consortium into the SNR process can further reduce the need for external aeration, as algae can provide O2 needed for aerobic heterotrophic bacteria (HB) and ammonia-oxidizing bacteria (AOB) (Wang et al. 2018). Wang et al. (2015) demonstrated that an algal–bacterial shortcut nitrogen removal (ABSNR) process could remove from AD sidestream with no external aeration other than mixing. The authors carried out the ABSNR process in a bench-scale photo-sequencing batch reactor (PSBR) with alternating light and dark periods (12 h light/12 h dark). During the light period, O2 was produced by algal photosynthesis, providing aerobic conditions for nitritation. During the dark period, the system became anoxic, providing favorable conditions for denitritation. Improved biomass production and nitritation efficiency was observed at higher light intensities. Arun et al. (2019) investigated the ABSNR process in a PSBR for the treatment of wastewater with C/N < 0.5 using a consortium of algae, AOB and methanol-consuming HB. Li et al. (2021) proposed a symbiotic algae-based SNR technology to reduce the cost of high
strength wastewater treatment. The authors used biofilm colonization in the algal–bacterial PSBR and studied the impacts of hydraulic retention time (HRT), aeration rate and carbon source addition on nitrogen removal efficiency.
All of the aforementioned ABSNR studies were carried out under artificial lighting with fixed light intensities during the light period (Wang et al. 2015, 2018; Arun et al. 2019; Li et al. 2021). However, full-scale photobioreactors, such as high-rate algal ponds (HRAPs), are usually operated under outdoor conditions, with day and night light cycles following seasonal patterns. During the summer, the high light intensity can increase DO concentrations to levels that may be toxic to algae and AOB (Nishi et al. 2020). While low light intensity during the winter or under overcast conditions will decrease photosynthetic O2 production and may require shallower reactors, supplemental illumination and/or longer HRT to maintain AOB growth. Operation of the PSBR at shallow depth or long HRT increases system footprint, land area requirements and capital costs. In addition, reduced nitrogen loading rates can reduce NOB suppression due to low FA concentrations (Duan et al. 2020). These challenges should be considered in PSBR design according to the influent wastewater composition and seasonal changes in light intensity. Therefore, mathematical models are needed to predict the efficiency of the ABSNR process under varying illumination similar to outdoor conditions and to address full-scale design challenges.
Several prior modeling studies have investigated algal–bacterial interactions and nitrogen removal pathways in ABSNR systems operated under fixed light intensities and dark periods (typically 12 h light/12 h dark). Arashiro et al. (2017) developed a mathematical model of the ABSNR process. The model was calibrated using data from a laboratory-scale ABSNR system operated with AD sidestream under varying operating conditions. The model described how total biomass concentration and solids retention time (SRT) affected light attenuation in the PSBR and consequently DO production and nitrogen removal. Peng et al. (2018) proposed a modeling framework for algal growth using a photo-inhibition parameter instead of mean light intensity. Synergistic and competitive growth kinetics were used to simulate algal–bacterial interactions in their study. Arun et al. (2019) also examined algal–bacterial interactions through metabolic models developed to predict nitrogen removal mechanisms based on algae-AOB and algae-AOB-HB activities. In a prior study by our group (Shayan et al. 2022), an ABSNR model was developed to predict the contributions of various nitrogen removal processes and the dominance of various biomass species during daily and long-term operating conditions (e.g., SRT, organic carbon requirements). The model was able to estimate the performance of laboratory-scale PSBRs fed with an AD sidestream under fixed light and dark periods with constant light intensity.
Solimeno & García (2019) developed the BIO-ALGAE model, which incorporated varying light conditions. The model simulated the dynamics of algae and bacteria in an HRAP over a year at different HRTs. The authors showed that the HRAP had different treatment capacities with changing seasons but did not propose a design to accommodate seasonal variations for annual operations. In addition, the BIO-ALGAE model did not include the NOB suppression required to simulate the SNR process for treatment of a high strength wastewater.
The aim of this study was to improve our previously developed model (Shayan et al. 2022) to include dynamics of algae, AOB, NOB and HB and dissolved chemical species (e.g., DO, organic and inorganic C, ,
,
) in ABSNRs operated with varying light intensities expected under outdoor operating conditions. The model was calibrated and validated using data from laboratory-scale PSBRs operated under light conditions mimicking winter (January 1990) and summer (June 1990) conditions in the Tampa Bay area (FL, USA), allowing us to test the performance under highly varying conditions. The model was used to investigate the impact of varying HRTs on algal–bacterial growth. Nitrogen removal performance was simulated for a full-scale PSBR used to treat AD sidestream wastewater in the Tampa Bay Area over an entire year using daily illumination data from 2020.
MATERIALS AND METHODS
Laboratory-scale PSBR design and operation
Photo-sequencing batch reactor (PSBR) schematic with details on operating cycles under varying illumination.
Photo-sequencing batch reactor (PSBR) schematic with details on operating cycles under varying illumination.
Alkalinity was adjusted at the beginning of each light period by adding MgCO3 to maintain the ratio of 3.4 mg CaCO3/mg -N for nitritation. Because the Mg2+ concentrations varied in the semi-synthetic centrate, the initial Mg2+ concentration varied from 65 to 150 mg/L. According to Daneshgar et al. (2018), MgCO3 addition can improve settling by facilitating the reaction between Mg2+,
and
to produce struvite, which precipitates in the pH range of 7–11. After nitritation in the light period and
generation, sodium acetate was added at the beginning of each dark period to initiate denitritation based on a stoichiometric molar ratio of acetate to
as 0.975:1.
Varying light intensity setup for laboratory-scale PSBR











Wastewater feed composition and operating conditions
A semi-synthetic AD sidestream was prepared using screened raw wastewater collected from the Falkenburg Advanced Wastewater Treatment Plant in Hillsborough County, Florida, with added NH4Cl, K2HPO4 and MgCO3 to achieve target concentrations summarized in Table 1. The semi-synthetic wastewater was formulated to mimic the characteristics of real sidestream from our prior studies, which was obtained by centrifuging biomass from a pilot AD-treating waste-activated sludge (Zalivina 2019). As mentioned previously, sodium acetate was added as the electron donor at the beginning of the dark period to facilitate denitritation. Summer mode was studied in Phase 1 at an HRT of 4 days, resulting in a nitrogen loading rate (NLR) of 87.5 mg N L−1 day−1. Winter mode was studied in Phase 2 at an HRT of 8 days, resulting in an NLR of 43.75 mg N L−1 day−1.
Operating parameters and average feed composition, for laboratory PSBRs
Operating parameters . | |
---|---|
Working volume | 2 L |
Hydraulic retention time (HRT) | 4-day (summer), 8-day (winter) |
Solids retention time (SRT) | 10-day |
Feed volume added per cycle Light intensity | 500 mL (summer), 250 mL (winter) Daily pattern (Figure S1) |
Operating pH range | 6.5–8.5 |
Operating alkalinity range | 400–700 mg CaCO3/L |
Feed composition | |
![]() | 350 ± 10 mg/L |
![]() | < 4 mg/L |
![]() | BDL |
![]() | 60 ± 5 mg/L |
Operating parameters . | |
---|---|
Working volume | 2 L |
Hydraulic retention time (HRT) | 4-day (summer), 8-day (winter) |
Solids retention time (SRT) | 10-day |
Feed volume added per cycle Light intensity | 500 mL (summer), 250 mL (winter) Daily pattern (Figure S1) |
Operating pH range | 6.5–8.5 |
Operating alkalinity range | 400–700 mg CaCO3/L |
Feed composition | |
![]() | 350 ± 10 mg/L |
![]() | < 4 mg/L |
![]() | BDL |
![]() | 60 ± 5 mg/L |
BDL, below detection limit.
Analytical methods
Grab samples were normally collected twice per week. Additional samples for model calibration and validation were collected during hourly sampling campaigns performed over a full cycle (24 h). Nitrogen species and phosphate concentrations were measured using a Metrohm Peak 850 Professional An/Cat-ion chromatography (IC) system (Metrohm Inc., Switzerland). DO and pH were measured in situ using calibrated Orion GS9156 meters (Thermo Fisher Scientific Inc., Waltham, MA, USA). Chlorophyll α was analyzed via NEN 6520-Dutch Standard. Samples were filtered through a 0.45 μm filter for soluble COD measurements using Lovibond COD test kits (Tintometer Inc., USA). rbCOD was measured using a rapid physical–chemical method (Mamais et al. 1993). Alkalinity was measured as CaCO3 using Standard Method 2320B. Total biomass density was measured as TSS using Standard Methods 2450B. Light intensity was measured by an ExTech Easyview 30 light meter (ExTech Inc., Waltham, MA, USA) at the liquid surface of the reactor.
Model description and reactor design
The model used the same structure as the algal–bacterial PSBR model presented in our prior study (Shayan et al. 2022), which combined parameters and process equations from Activated Sludge Model No.1 (ASM1), River Water Quality Model No.1 (RWQM1) and BIO-ALGAE (Henze et al. 2000; Reichert et al. 2001; Solimeno et al. 2017a; Solimeno & García 2019). The model includes 17 dissolved and particulate species and 24 physical, chemical, and biochemical process rates (Table S1). Furthermore, a matrix of stoichiometric coefficients and their values were listed in Tables S2 and S3 in the supplementary information. Table S4 also included the biokinetic, chemical and physical parameters in the model. No biological processes were considered during the last 2 h of settling and decanting. Effluent characteristics from each cycle were used as influent values for the next cycle. Since the PSBRs were operated under LED lights that mimicked outdoor seasonal light patterns, the model was modified to use light intensity as a function of time. In addition, the sensitive kinetic parameters listed in the prior study were re-evaluated and calibrated manually within the ranges in the literature to minimize the errors between the model predictions and the experimental results according to changes in operating conditions. The calibrated model was validated using data from hourly sampling campaigns carried out on PSBRs operated in summer and winter modes.
The validated model was used to simulate a full-scale PSBR assumed to be used for the treatment of AD sidestream generated at the South Cross Bayou (SCB) Water Reclamation Facility in Pinellas County, Florida. The case study was suggested in our preliminary investigation (Shayan 2021), with typical AD sidestream conditions for SCB reported by Medina (2020) and summarized in Table 2. Ten-day moving averages of daily light irradiances for the year 2020 from a nearby monitoring station were used. Global solar irradiance is usually measured in W/m2 and an approximate conversion factor of 4.6 is typically used to estimate light intensity in μmol/m2 s (Carruthers et al. 2001). However, since ∼ 45% of solar radiation falls in the photosynthetic active range, the conversion factor used in this study was 1 W/m2 = 2.1 μmol/m2 s.
Influent wastewater characteristics in the full-scale model (Medina 2020)
Parameter . | Influent wastewater . |
---|---|
Flow rate | 557 m3/day |
![]() | 715 mg/L |
![]() | 1 mg/L |
![]() | 0.1 mg/L |
HB (XH) | 150 mgCOD/L |
AOB (XAOB) | 50 mgCOD/L |
NOB (XNOB) | 1 mgCOD/L |
Alg (XP) | 2 mgCOD/L |
Inert particulate organic matter (XI) | 20 mgCOD/L |
Slowly biodegradable organic matter (XS) | 20 mgCOD/L |
Readily biodegradable substrate (SS) | 50 mgCOD/L |
pH | 7.5 |
Alkalinity | 850 mgCaCO3/L |
Parameter . | Influent wastewater . |
---|---|
Flow rate | 557 m3/day |
![]() | 715 mg/L |
![]() | 1 mg/L |
![]() | 0.1 mg/L |
HB (XH) | 150 mgCOD/L |
AOB (XAOB) | 50 mgCOD/L |
NOB (XNOB) | 1 mgCOD/L |
Alg (XP) | 2 mgCOD/L |
Inert particulate organic matter (XI) | 20 mgCOD/L |
Slowly biodegradable organic matter (XS) | 20 mgCOD/L |
Readily biodegradable substrate (SS) | 50 mgCOD/L |
pH | 7.5 |
Alkalinity | 850 mgCaCO3/L |
RESULTS AND DISCUSSION
Laboratory-scale PSBR operation
Laboratory-scale PSBRs were operated under summer and winter illumination conditions using the converted light intensities and influent wastewater concentrations listed in Table 1. Under each condition (summer/winter) the PSBRs were operated for at least 30 daily cycles after a steady state was reached. Weekly samples from specific times (start, mid, end) were taken to calculate means and standard deviations (Tables 3 and 4).
Mean values and standard deviations (SD) of oxygen and nitrogen species from weekly samples at specific times under summer conditions
Time (day) . | 0.04 . | 0.25 . | 0.9 . | |||
---|---|---|---|---|---|---|
Mean . | SD . | Mean . | SD . | Mean . | SD . | |
O2 (mg/L) | 0.11 | 0.05 | 0.35 | 0.04 | 0.45 | 0.19 |
![]() | 56.33 | 2.25 | 30.33 | 1.75 | 0.14 | 0.23 |
NO2-N (mg/L) | 0.07 | 0.08 | 5.75 | 0.44 | 2.85 | 0.22 |
NO3-N (mg/L) | 0.03 | 0.05 | 0.02 | 0.04 | 0.005 | 0.008 |
Time (day) . | 0.04 . | 0.25 . | 0.9 . | |||
---|---|---|---|---|---|---|
Mean . | SD . | Mean . | SD . | Mean . | SD . | |
O2 (mg/L) | 0.11 | 0.05 | 0.35 | 0.04 | 0.45 | 0.19 |
![]() | 56.33 | 2.25 | 30.33 | 1.75 | 0.14 | 0.23 |
NO2-N (mg/L) | 0.07 | 0.08 | 5.75 | 0.44 | 2.85 | 0.22 |
NO3-N (mg/L) | 0.03 | 0.05 | 0.02 | 0.04 | 0.005 | 0.008 |
Mean values and standard deviations (SD) of oxygen and nitrogen species from weekly samples at specific times under winter conditions
Time (day) . | 0.02 . | 0.25 . | 0.9 . | |||
---|---|---|---|---|---|---|
Mean . | SD . | Mean . | SD . | Mean . | SD . | |
O2 (mg/L) | 0.10 | 0.05 | 0.43 | 0.05 | 0.005 | 0.008 |
![]() | 55.31 | 3.21 | 27.24 | 1.86 | 0.21 | 0.23 |
NO2-N (mg/L) | 9.34 | 0.98 | 22.26 | 2.06 | 4.67 | 0.42 |
NO3-N (mg/L) | 1.12 | 0.44 | 1.61 | 0.08 | 1.02 | 0.075 |
Time (day) . | 0.02 . | 0.25 . | 0.9 . | |||
---|---|---|---|---|---|---|
Mean . | SD . | Mean . | SD . | Mean . | SD . | |
O2 (mg/L) | 0.10 | 0.05 | 0.43 | 0.05 | 0.005 | 0.008 |
![]() | 55.31 | 3.21 | 27.24 | 1.86 | 0.21 | 0.23 |
NO2-N (mg/L) | 9.34 | 0.98 | 22.26 | 2.06 | 4.67 | 0.42 |
NO3-N (mg/L) | 1.12 | 0.44 | 1.61 | 0.08 | 1.02 | 0.075 |


Dynamic simulation of 24 h summer mode operation with 4-day HRT and using the kinetic parameters obtained for constant light/dark mode (green line) and recalibrated parameters for the varying light condition (blue line).
Dynamic simulation of 24 h summer mode operation with 4-day HRT and using the kinetic parameters obtained for constant light/dark mode (green line) and recalibrated parameters for the varying light condition (blue line).
The adaptations of the biomass to outdoor light conditions and resulting changes in nitrogen removal efficiency justified the recalibration of the model for varying light intensity conditions. Table 5 shows the sensitive parameters and adjusted values compared with values obtained for the constant light intensity study. Simulation results using the recalibrated kinetic parameters are shown in Figure 2. Sensitivity analysis was carried out by manually varying the kinetic parameters (±10%) and checking the target concentration response. This was the same method used in a previous study under constant illumination (Shayan 2021). As an example, Figure S2 in the supplementary information shows O2 variations with different bH values (0.35–0.55 day−1). Sensitive parameters were adjusted and calibrated manually and individually based on experimental conditions and reported values or ranges in similar studies (Table S4 in supplementary information). The primary reason for some significant changes is switching the PSBR operation from constant light into varying light intensities. Since biomass growth, nitrogen removal and oxygen generation were adapted for new operating conditions under long-term operation, some kinetic parameters needed to be significantly adjusted. For example, the ammonium inhibition constant, was decreased from 65.5 to 7.5 mg/L to reflect greater NOB inhibition under varying light intensity and get closer to the value reported by Solimeno et al. (2017b) for an outdoor HRAP. The half saturation constant of HB for
was decreased to account for
consumption by HB during the light period and simultaneous nitritation/denitritation. The half saturation constant of HB for
(
was increased to describe
accumulation under varying light intensity without rapid denitritation. Furthermore, to reflect low HB growth at early light periods, the saturation constant for rbCOD (
) was increased significantly to minimize HB growth in early light period and competition with AOB. This value was also reported by Solimeno et al. (2017b) for outdoor HRAP operations. The half saturation constant for algae growth on
(
) was decreased to reflect the algae growth under low
concentrations when there is higher light intensity in daytime. This can promote photosynthesis, algae growth and O2 generation by increased illumination. On the other hand,
and
were increased to reflect the lower AOB growth rate during low O2 availability periods at the beginning of the cycle compared with the constant light/dark model (Shayan et al. 2022).
Sensitive parameters for model based on gradual light illumination and their values for both constant and variable light illumination
Kinetic parameters . | Constant light illumination (Shayan et al. 2022) . | Variable light illumination . | Unit . |
---|---|---|---|
![]() | 0.35 | 0.5 | day−1 |
![]() | 0.2 | 0.1 | day−1 |
![]() | 65.5 | 7.5 | mg/L |
![]() | 0.8 | 1.7 | mg/L |
![]() | 1.5 | 0.5 | mg/L |
![]() | 0.5 | 6.5 | mg/L |
![]() | 3 | 1 | mg/L |
![]() | 0.028 | 0.1 | mg/L |
![]() | 0.85 | 0.8 | mg/L |
![]() | 0.3 | 0.81 | mg/L |
![]() | 5 | 20 | mg/L |
Kinetic parameters . | Constant light illumination (Shayan et al. 2022) . | Variable light illumination . | Unit . |
---|---|---|---|
![]() | 0.35 | 0.5 | day−1 |
![]() | 0.2 | 0.1 | day−1 |
![]() | 65.5 | 7.5 | mg/L |
![]() | 0.8 | 1.7 | mg/L |
![]() | 1.5 | 0.5 | mg/L |
![]() | 0.5 | 6.5 | mg/L |
![]() | 3 | 1 | mg/L |
![]() | 0.028 | 0.1 | mg/L |
![]() | 0.85 | 0.8 | mg/L |
![]() | 0.3 | 0.81 | mg/L |
![]() | 5 | 20 | mg/L |
The recalibrated model predictions generally matched well with the measured concentrations of the dissolved species, as shown in Figure 2. The model was able to simulate O2 consumption during 0 < t < 0.1 day due to AOB and HB growth, and O2 accumulation after declining concentration and decreasing AOB growth (Figure 2(a)). Nitrogen species diagrams showed
oxidation and
production during the daytime (Figure 2(b)) and denitritation during the night (Figure 2(c)). Figure 2(d) showed that a negligible amount of
was produced over the complete cycle, which confirmed the SNR process. Increases in O2 concentrations at the end of the cycle (0.9 < t < 1 day) were most likely due to disturbances of the PSBR by sampling the biomass during the settling phase. In addition to R2, a comparison of the root mean square error (RMSE) and normalized RMSE (NRMSE) between the simulation results from constant and varying light kinetic parameters, showed the importance of recalibration for studies under varying light intensities (Table 6).
Comparison of R2, root mean square error (RMSE) and normalized RMSE (NRMSE) PSBR simulations using kinetic parameters from constant light and recalibrated parameters for varying light
. | O2 . | ![]() . | ![]() . | ![]() . | ||||
---|---|---|---|---|---|---|---|---|
Illumination . | Varying . | Constant . | Varying . | Constant . | Varying . | Constant . | Varying . | Constant . |
R2 | 0.81 | 0.03 | 0.98 | 0.98 | 0.86 | 0.76 | 0.98 | 0.002 |
root mean square error (RSME) (mg/L) | 0.38 | 0.88 | 1 | 3.37 | 1.37 | 2.69 | 0.001 | 0.001 |
normalized root mean square error (NRSME) | 0.71 | 1.62 | 0.06 | 0.19 | 0.22 | 0.42 | 0.05 | 0.73 |
. | O2 . | ![]() . | ![]() . | ![]() . | ||||
---|---|---|---|---|---|---|---|---|
Illumination . | Varying . | Constant . | Varying . | Constant . | Varying . | Constant . | Varying . | Constant . |
R2 | 0.81 | 0.03 | 0.98 | 0.98 | 0.86 | 0.76 | 0.98 | 0.002 |
root mean square error (RSME) (mg/L) | 0.38 | 0.88 | 1 | 3.37 | 1.37 | 2.69 | 0.001 | 0.001 |
normalized root mean square error (NRSME) | 0.71 | 1.62 | 0.06 | 0.19 | 0.22 | 0.42 | 0.05 | 0.73 |



Conversion rate of O2 in a complete cycle of PSBR in summer illumination model with HRT 4 days.
Conversion rate of O2 in a complete cycle of PSBR in summer illumination model with HRT 4 days.






Goodness of fit and statistical errors from model validation by summer light illumination and HRT of 4 days
. | O2 . | ![]() | ![]() | ![]() |
---|---|---|---|---|
R2 | 0.83 | 0.98 | 0.79 | 0.98 |
RMSE (mg/L) | 0.04 | 3.63 | 0.09 | 0.001 |
NRMSE | 0.33 | 0.08 | 0.49 | 0.09 |
. | O2 . | ![]() | ![]() | ![]() |
---|---|---|---|---|
R2 | 0.83 | 0.98 | 0.79 | 0.98 |
RMSE (mg/L) | 0.04 | 3.63 | 0.09 | 0.001 |
NRMSE | 0.33 | 0.08 | 0.49 | 0.09 |
Experimental results and model validation for a complete cycle of PSBR trial 2 under the summer light illumination and HRT of 4 days.
Experimental results and model validation for a complete cycle of PSBR trial 2 under the summer light illumination and HRT of 4 days.





Goodness of fit and statistical errors from model validation by winter light illumination and prolonged HRT (8 days)
. | O2 . | ![]() | ![]() | ![]() |
---|---|---|---|---|
R2 | 0.78 | 0.96 | 0.91 | 0.81 |
RMSE (mg/L) | 0.09 | 3.87 | 3.41 | 0.51 |
NRMSE | 0.57 | 0.15 | 0.21 | 0.38 |
. | O2 . | ![]() | ![]() | ![]() |
---|---|---|---|---|
R2 | 0.78 | 0.96 | 0.91 | 0.81 |
RMSE (mg/L) | 0.09 | 3.87 | 3.41 | 0.51 |
NRMSE | 0.57 | 0.15 | 0.21 | 0.38 |
Experimental results and model validation for a complete cycle of PSBR under the winter light illumination and prolonged HRT (8 days).
Experimental results and model validation for a complete cycle of PSBR under the winter light illumination and prolonged HRT (8 days).
Impacts of HRT on PSBR operation
To investigate the impact of HRT on biomass species distribution at a constant SRT of 10 days, the model was used to simulate long-term PSBR operation under summer and winter illumination conditions at HRTs of 2, 4, 6 and 8 days (Table 9). Simulation results for 120 days of PSBR operation after the acclimation phase under summer and winter conditions showed the effect of HRT on different biomass species in the model. All microbial species concentrations reached steady state within 30–50 days of simulation. The steady state and simulated concentrations of HB, AOB and NOB decreased with increasing HRT in both summer and winter modes. Longer HRT will decrease substrate loading rate (e.g., ammonia), resulting in lower biomass growth and therefore lower biomass concentrations if constant SRT is maintained.
For HB, the simulated concentrations under steady-state winter mode were higher than that in summer mode. This could be the result of lower light intensity, lower availability of O2 and more favorable conditions for anoxic growth of HB in winter than summer. For AOB and NOB, the steady state concentrations were generally higher for summer mode compared with winter mode. This is because AOB and NOB growth are highly dependent on O2 availability. Higher light intensities in summer mode than in winter mode resulted in greater photosynthesis rates and oxygen generation by algae growth. For algae, the impact of HRT in winter mode was different from that in summer mode. Winter mode simulations for algae showed the minimum steady state concentration at an HRT of 2 days. This could be due to higher biomass density of HB, AOB and NOB in winter mode with an HRT of 2 days, which reduces light availability in the medium and decreases algae growth.
Steady state concentration of biomass species during 120 days of PSBR simulation in summer and winter illuminations with different HRTs
. | HRT 2 days . | HRT 4 days . | HRT 6 days . | HRT 8 days . |
---|---|---|---|---|
Algae (mg/L) | ||||
Summer | 2,200 | 1,550 | 1,500 | 1,500 |
Winter | 450 | 510 | 510 | 510 |
Heterotrophic bacteria (mg/L) | ||||
Summer | 160 | 120 | 115 | 110 |
Winter | 250 | 225 | 200 | 165 |
AOB (mg/L) | ||||
Summer | 45 | 20 | 12 | 8 |
Winter | 25 | 18 | 16 | 10 |
NOB (mg/L) | ||||
Summer | 0.45 | 0.35 | 0.3 | 0.25 |
Winter | 0.55 | 0.3 | 0.2 | 0.15 |
. | HRT 2 days . | HRT 4 days . | HRT 6 days . | HRT 8 days . |
---|---|---|---|---|
Algae (mg/L) | ||||
Summer | 2,200 | 1,550 | 1,500 | 1,500 |
Winter | 450 | 510 | 510 | 510 |
Heterotrophic bacteria (mg/L) | ||||
Summer | 160 | 120 | 115 | 110 |
Winter | 250 | 225 | 200 | 165 |
AOB (mg/L) | ||||
Summer | 45 | 20 | 12 | 8 |
Winter | 25 | 18 | 16 | 10 |
NOB (mg/L) | ||||
Summer | 0.45 | 0.35 | 0.3 | 0.25 |
Winter | 0.55 | 0.3 | 0.2 | 0.15 |
A shorter HRT can lead to higher AOB concentrations as discussed previously; however, higher ammonia loading rates can result in high FA concentrations that can reach the FA inhibition threshold for AOB (10–150 mg/L) (Soliman & Eldyasti 2016) at the beginning of the cycle. The calculated FA concentration for an HRT of 4 days based on the /NH3 equilibrium showed that FA was in the acceptable range (0.1–3.5 mg/L) to inhibit NOB growth. Also, the calculated FNA concentration for an HRT > 4 days was ∼0.005 mg/L. Therefore, a longer HRT will favor NOB suppression; however, it will require a larger reactor, which may be costly. Since algae and AOB growth showed almost the same trends during HRTs of 4, 6 and 8 days, an HRT of 6 days was selected for year-round illumination conditions for a full-scale PSBR design considering the tradeoff between promoting NOB suppression and minimizing reactor volume.
Large-scale PSBR simulation
Sidestream characteristics from the SCB treatment plant in Pinellas County, FL were based on the modeling results from Medina (2020) using BioWin software (EnviroSim, Ontario, Canada). Effluent characteristics from dewatering simulations and the plant's available reports (Table 2) were used for the large-scale PSBR, assuming ideal mixing conditions. The assumed SRT and HRT for the simulations were 10 and 6 days, respectively.








Normalized concentrations of dissolved and particulate species during long-term operation of an outdoor assumed large-scale PSBR under HRT of 6 days.
Normalized concentrations of dissolved and particulate species during long-term operation of an outdoor assumed large-scale PSBR under HRT of 6 days.
Fluctuations in some chemical species concentrations during long-term simulation resulted from seasonal light intensity variations (e.g., cloudy conditions) (Figure 6(a)). For example, effluent -N concentration of the assumed PSBR with a depth of 1.2 m, varied between 0 and 88 mg/L for a simulation period of 380 days, while the mean concentration was < 50 mg/L. The dynamics of O2 and nitrogen species concentrations in a full-scale simulation for two selected days (May 15 and November 15) were predicted and showed that nitrogen removal is higher during the summer than in winter due to greater light intensities and longer illumination periods in a complete cycle (see Figure S6 in supplementary information). In addition, pH variations were simulated for the selected days of the above-mentioned reactor and trends were similar to experimental results on a laboratory scale for summer and winter modes (Figure S7).
While ABSNR simulation for the assumed PSBR showed sufficient nitrogen removal under seasonal light variations, decreasing the reactor depth can increase light penetration to minimize concentration fluctuations over the year. Decreased reactor depth will likely result in increased land requirements and capital costs.
Recommendations for future research
The model developed in this study can be used as a tool to estimate algal–bacterial interactions in an ABSNR process under seasonal changes in light intensity. A number of assumptions were made to simplify the mathematical model, particularly with regard to the full-scale system simulation. First, we assumed ideal mixing such that dissolved and particulate matter were homogenously distributed inside the reactor. Non-ideal hydrodynamics in a real photobioreactor, such as HRAPs, results in the development of dead zones (i.e., where anoxic processes can occur), short circuiting, and affect mass transfer rate coefficients and fluxes of gases (e.g., O2, FA) to the atmosphere (Hadiyanto et al. 2013). Second, we assumed a constant temperature of 20 °C; however, the temperature is known to fluctuate in outdoor algal ponds that can impact algal growth and productivity (Vindel & Trincado 2021). Third, the assumed PSBR depth was > 0.5 m, which is uncommon in real designs and only used to estimate the nitrogen removal capacity for a deep reactor with limited surface area. In practice, a range of 0.1–0.3 m has been recommended for this purpose (Hadiyanto et al. 2013). Future studies should consider shallower ABSNR designs, which would allow for greater illumination of the algal biomass. Fourth, our model only included assimilation, nitrification, denitrification and FA volatilization as major N transformation processes; however, the formation of nitrous oxide (N2O) as a byproduct of bacterial and algal metabolism should not be neglected (Zhang et al. 2022). Future studies should investigate N2O generation in ABSNR processes. Lastly, our model assumed that sodium acetate was the carbon source for denitrification based on our laboratory-scale PSBR studies. Future studies should consider other denitritation electron donors, such as methanol, ethanol or primary sludge fermentate, and the effects of these chemicals on biodegradation rates.
CONCLUSIONS
In this study, a mathematical model was developed to simulate the ABSNR process under seasonal light irradiance conditions. Data from laboratory-scale PSBR studies were used for model calibration and validation. The dynamics of microbial biomass, DO, rbCOD and dissolved nitrogen species concentrations were examined over 380 days of PSBR simulation with seasonal light variations. The model was used to identify an appropriate HRT and depth for a PSBR and to elucidate the effect of seasonal light availability on nitrogen removal performance. A full-scale PSBR could achieve >90% ammonia removal, significantly reducing the nitrogen load to the mainstream wastewater treatment plant. The dynamic simulation showed that the ABSNR process can help wastewater treatment facilities meet stringent nitrogen removal standards with low energy inputs.
ACKNOWLEDGEMENTS
This research was supported by the US National Science Foundation (NSF) under Grant No. 1511439.
DISCLOSURE STATEMENT
The authors report no commercial or proprietary interest in any product or concept discussed in this article.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.