Effect of Hydraulic Retention Time on Nutrient Removal in a Microalgae-Based Tertiary Treatment: A Pilot-Scale Study in Winter Conditions

Abstract

The wastewater treatment (WWT) industry is currently facing challenges imposed by the revised urban WWT directive, particularly in terms of nitrogen (N) and phosphorus (P) removal. This implies the need for mandatory tertiary treatment, for which microalgae cultivation shows great sustainability promise. This study investigated the impact of hydraulic retention time (HRT) on nutrient removal in open-air microalgae cultivation for tertiary WWT under winter conditions. Two pilot-scale semi-continuous raceway systems were operated with indigenous microalgae, natural sunlight, and no pH control. HRT values of 4, 5.5, and 7 days were tested, and N, P, and carbon (C) removal and recovery were measured. All conditions allowed nitrogen removal, complying with the revised urban WWT directive. Regarding P, only the 7-day HRT condition consistently complied with the directive’s lowest limit (<0.5 mg P·L⁻¹) in the treated water, while 5.5 and 4 days left up to 0.7 and 1.0 mg P·L⁻¹, respectively, in up to 25% of the samples. A stable microalgae consortium was established under variable light, pH, and dissolved oxygen conditions, albeit with variable biomass productivity. Elemental mass balances revealed that nutrients were mostly recovered in the produced biomass, particularly at high HRT, including effective CO₂ capture from the atmosphere.

1. Introduction

To reduce the human impact on water bodies, the European Union (EU) has been tightening the demands on wastewater management and discharge [1]. However, conventional wastewater treatment (WWT) with limited recovery of materials and the remediation of water quality as the main goal is not sustainable [2]. Therefore, the implementation of the circular economy concept in this sector has been a focus point for the research community [3]. In conventional wastewater treatment processes, throughout the stages from preliminary to tertiary, priority is given to pollutant removal efficacy, and the valorisation of the removed pollutant load is only addressed in the sludge treatment stage [4]. The addition to this conventional treatment of processes that allow further recovery of valuable components is therefore required [3,5].

Nitrogen and phosphorus are abundant in urban wastewater and are the main cause of eutrophication in natural water bodies, with the proliferation of algal blooms feeding on the increased load of nutrients. These blooms disturb the natural ecosystem balance and may cause the loss of species diversity [6]. Thus, the new EU directive regarding urban wastewater (UWW) treatment and discharge is setting new targets for nitrogen (<8 mg N·L⁻¹) and phosphorous (<0.5 mg P·L⁻¹) discharge limits and the mandatory implementation of tertiary treatment [1]. Controlled microalgae cultivation at the tertiary treatment stage is a promising way to accommodate these new regulatory limits, with nitrogen and phosphorus being readily removed from the water for microalgae growth [7,8]. Moreover, microalgae processes produce biomass that can be further valorised and promote CO₂ sequestration from the atmosphere, contributing to carbon neutrality [8,9,10].

Raceway (RW) bioreactors are widely used for open-air microalgae cultivation, and several works have been published on wastewater treatment in these systems, namely for primary [11,12,13] and secondary effluents [14,15,16] and to treat liquid digestates from anaerobic digestion of wastewater sludge [17,18]. Operation in continuous or semi-continuous modes at hydraulic retention time (HRT) values between 4 and 10 days were the typical cases.

A number of reviews have been recently published on the challenge of integrating microalgae cultivation in wastewater treatment systems [19,20,21,22,23]. Nevertheless, published studies on pilot-scale microalgae cultivation in RW or similar continuous stirred pond systems for tertiary treatment, i.e., fed with secondary effluent, are scarce, particularly under low-complexity conditions. Studies at this scale using primary effluent, centrate, digestate, or other industrial effluents are more abundant [10]. Also, some of the tertiary treatment studies involved operational features of increased complexity, such as aeration [24] and CO₂ supplementation for pH control [16,25], weather-protection covers [15], artificial illumination [15], inoculation with specific algae species [14,15], tap water addition [14], and nutrient supplementation [26]. Treatment effectiveness was mostly assessed through the removal yields of nitrate (N-NO₃⁻), ammonia nitrogen (N-NH₄⁺), total nitrogen (TN), phosphate (P-PO₄³⁻), total phosphorus (TP), and chemical oxygen demand (COD). However, most studies monitored only a selection of these components, and elemental mass balances to determine nutrient fate are not presented. Assumptions, such as ammonia volatilisation, are sometimes made based on operational conditions like culture pH, but no identification and quantification of the dominant removal mechanisms are performed, namely the fraction which is incorporated in the produced biomass [27]. Calculating mass balances on nitrogen, phosphorus, and carbon is a key factor to determine if the treatment is meeting its circularity objectives, with the nutrients being removed from the water and recovered in the biomass for further valorisation, instead of being released into the atmosphere [5].

In the present study, the use of microalgae cultivation for the tertiary treatment of UWW was assessed, focusing on low-complexity, open-air operation at pilot scale, with no supplementation or inoculation, a natural microalgae consortium being allowed to establish itself. The feed was the effluent from the secondary stage of a wastewater treatment plant receiving both domestic and industrial effluents and performing advanced nitrogen removal. Two RW systems were operated in parallel, with different HRT values, to identify the minimum value leading to nutrient limitation, while covering variations in weather conditions. The tests were performed throughout the 2023/24 winter season in Portugal, which is the harshest season for microalgae cultivation due to low solar irradiation and temperature values and frequent rainfall periods. The system’s resilience to the occurrence of high pH and dissolved oxygen concentration periods, enhanced by RW operation in a semi-continuous regime, was also evaluated, namely in what concerned the microalgae types present and contaminant microfauna (grazer) abundance. Biomass growth and nutrient removal rates were calculated, and mass balances were performed to quantify the dominant removal routes for the carbon, nitrogen, and phosphorus inputs in the treatment system. Hence, the main objective of this study is to determine the feasibility of introducing a microalgae cultivation process as a tertiary step for enhanced N and P removal in unfavourable environmental conditions, aiming at full-scale implementation.

2. Materials and Methods

2.1. Experimental Set-Up

Tests were performed in two pilot-scale conventional raceway systems (Figure 1) located at a WWT plant in Portugal during the winter season, from 27 December 2023 to 25 March 2024. Each raceway unit had outer dimensions of 5 m in length, 1.1 m in width, 1.1 m in height not including the paddlewheel, and 1.5 m in total height. Mixing was achieved through a paddlewheel located in each system (1.7 m from the nearest end), at 12 rpm rotation frequency. The working volume of each RW was 700 L, and the photosynthetic area was 5 m². The water depth after each feeding was set at 16 cm, and it fluctuated between 14.8 and 21.4 cm between feeds due to evaporation and precipitation. Secondary effluent diverted from the outlet of the WWT plant’s secondary settler, after activated sludge treatment with advanced nitrogen removal, was used as feed throughout the experiment. This effluent’s characterisation is described in

Table 1. Secondary effluent characterisation. Average and standard deviation calculated excluding outliers (see Section 2.5). Outliers are included in the min–max ranges. Number of samples analysed: 34. TIC—total inorganic carbon; TOC—total organic carbon; TC—total carbon; LQ limit of quantification of the analytical method.

Nutrient	Average (mg·L−1)	Standard Deviation (mg·L−1)	Min (mg·L−1)	Max (mg·L−1)	LQ (mg·L−1)
N-NO3−	5.2	1.8	0.2	8.6	0.06
N-NH4+	5.0	4.1	0.8	31.3	0.78
TN	9.4	4.1	1.7	32.0	0.005
P-PO43−	2.1	1.3	0.2	26.3	0.2
TP	2.3	1.3	0.4	32.4	0.18
TIC	22.0	6.9	9.7	41.8	0.005
TOC	6.6	1.8	2.5	14.6	0.005
TC	28.5	8.4	12.2	56.4	0.005

. Whenever values below the limit of quantification (LQ) of the analytical method were reached, the value assumed was the LQ itself.

2.2. Operational Conditions

To allow the development of a spontaneous microalgae population, a preliminary operation stage was carried out, with the duration of 13 weeks. During this, both RW systems were filled with secondary effluent and operated at an HRT of 7 days, without initial inoculation, in semi-continuous, fill-and-draw mode. Subsequently and throughout this study, the systems were operated in the same fill-and-draw mode, with renewal cycles every 3–5 days, depending on the set HRT value. At the start of each cycle, part of the RW contents was discarded, the remaining volume being kept as inoculum for the following cycle, and fresh secondary effluent was added. The effluent volume to be added was calculated from the set HRT value and the set cycle time as

$$V_{added}={\displaystyle \frac{V}{HRT/∆t}},$$

(1)

where V_added is the volume of fresh effluent added, V is the working volume of the reactor (700 L), and Δt is the cycle time. The culture volume retained in the RW, V_retained, that acted as inoculum was calculated as

$$V_{retained}=V-V_{added}.$$

(2)

The efficacy of the wastewater treatment was assessed for each renewal cycle, i.e., the time between each fresh effluent addition, hereinafter named the treatment cycle, operated under each condition.

Liquid volumes inside the RWs were measured with a precision of 25 L using a vertical scale set vertically against the RW’s inner wall, previously calibrated for each RW with known added volumes of water. At the end of each cycle, before sampling and contents removal, homogenisation of the liquid was performed with a manual paddle fitted with a rubber scraper: The aim was to dislodge biomass that accumulated on the walls of the reactor and incorporate it in the mixed suspension in the form of biomass aggregates of reduced size, allowing representative sampling. The culture samples were always collected at mid-depth at the same location in each RW, immediately downstream of the paddlewheel, to provide representative and comparable aliquots.

The impact of the HRT on wastewater treatment effectiveness and biomass productivity was assessed, with values from the 4–10 day range found in the literature (see Section 1). An HRT value of 7 days was selected as the control setting, since during the microalgae culture development stage it repeatedly led to nutrient depletion. Also, due to this, only lower HRT values were selected as test conditions. Thus, the HRT values tested were circa 7, 5.5, and 4 days. One RW (RW1) was always operated at an HRT of 7 days, while the other RW (RW2) alternated between cycle sequences at 4 and 5.5 days (

Table 2. Operational HRT conditions tested. The 7-day HRT condition was operated in RW 1 and the 5.5- and 4-day HRT conditions were operated in RW2.

Set HRT (Days)	Measured HRT (Days)	Dates	Operating Days	Number of Cycles
7	6.90 ± 0.08	27 December 2023–25 March 2024	89	30
5.5	5.51 ± 0.06	26 January 2024–6 March 2024	40	14
4	3.92 ± 0.03	5 January 2024–24 January 2024 6 March 2024–22 March 2024	35	15

).

2.3. Sampling and Analytical Procedures

At the beginning of each renewal cycle, 1 L samples were collected from the secondary effluent to be fed to the RWs (effluent samples) and from the mixed suspension inside the RWs immediately after effluent feeding (beginning samples). At the end of each cycle, 1 L samples were collected from the homogenised contents of each RW before discharge (end samples). All samples were processed within 4–5 h of collection. For solute analysis, samples were first clarified through centrifugation at 3500 rpm for 10 min (centrifuge HERMLE Z 400 K, Wehingen, Germany), and then stored at −20 °C until further processing.

Biomass growth was assessed through dry cell weight (DW) measurements using the total suspended solids method (method 2540-D) [28]. Filters of 0.7 µm pore size were used, and the filters with samples were dried in a moisture analyser (AND MS-70, Tokyo, Japan) at 180 °C. There is no distinction between algal biomass and other microbial biomass or suspended solids retained on the filter.

The microalgae consortium and contaminants present were examined through optical microscopy (OLYMPUS BX53F, Tokyo, Japan) on samples of the discharged RW contents. Microalgae and contaminant (grazer) identification was qualitative, achieved only through morphology observation and comparison with literature images [28,29]. Photographs of the microorganisms are presented as Supplementary Materials to ensure reproducibility of the identification. The relative abundance of each microalgae was expressed through semi-quantitative levels at intervals ]0–5]%, ]5–20]%, ]20–40]%, ]40–60]%, ]60–80]%, and ]80–100]%. For the contaminants, a level from 1 to 5 was attributed to each, where 1 corresponds to minimum abundance and 5 to high abundance, and their total abundance was subsequently calculated as the sum of the individual levels for all contaminants present. Similarly, semi-quantitative approaches have been used in other studies, mainly to identify significant shifts in the culture composition instead of identifying the specific abundance of each microorganism [30].

Nitrate (N-NO₃⁻) concentration was measured using the ultraviolet spectrophotometric screening method (method 4500-NO3--B) [28]. Ammonia nitrogen (N-NH₄⁺) was measured using the phenate method (method 4500-NH3-F) [28], phosphate (P-PO₄³⁻) was measured using the ascorbic acid method (method 4500-P-E) [28], and total phosphorus (TP) was measured using the persulfate method (method 4500-P-J) [28] followed by phosphate quantification. Total nitrogen (TN) was measured using the thermal decomposition/chemiluminescence methods; total carbon (TC), total inorganic carbon (TIC), and total organic carbon (TOC) were measured using the high-temperature combustion method (method 5310-B) [28]. A TOC/TN automatic analyser (Shimadzu, Kyoto, Japan) was used for all these determinations, and the procedures were carried out according to the manufacturer’s indications.

The biomass pellets from sample centrifugation were washed with distilled water to ensure that their elemental analysis was not affected by the presence of precipitates or residues of supernatant. Biomass nitrogen and carbon contents were determined through elemental analysis (CHNS) using NC technologies equipment (ECS 8040 CHNS-O, Milan, Italy), and the phosphorus content was determined using the TP method described above.

For all spectrophotometric measurements a microplate reader (SPECTROstar Nano, Ortenberg, Germany) was used, and calibration curves (presented in Supplementary Materials) were obtained previously for the concerned analytes. For the ammonia nitrogen and phosphate method kits with pre-prepared reagents were used (SERA, Heinsberg, Germany).

Online, in situ measurements of temperature, pH, dissolved oxygen and oxidation reduction potential were performed at 15 min intervals with a multiparametric probe inserted into the liquid in RW2 (HANNA HI98494, Cluj-Napoca, Romania). Environmental data (irradiance and temperature) were collected by a meteorological station (SAINLOGIC WS3500, Shenzhen, China) located at the wastewater treatment facility. PAR irradiance values were calculated from the solar irradiance using the conversion factor of 2.02 µmol·m⁻²·s⁻¹ per 1 W·m⁻² [31].

2.4. Performance Indicators and Mass Balances

The efficacy of the wastewater treatment and biomass production was assessed for each renewal cycle performed at each condition tested. All mass values (g) used in these calculations were determined by multiplying measured values of concentration (Section 2.3) and volume (Section 2.2).

The biomass growth productivity was calculated as

$$Productivity={\displaystyle \frac{{DW}_{end}-{DW}_{beginning}}{∆t\times A}},$$

(3)

where DW_beginning is the biomass (g) present in the RW at the beginning of the cycle, DW_end is the biomass (g) present in the RW at the end of the cycle, Δt is the duration of the cycle (days) and A is the photosynthetic area of the RW (m²).

The removal rate for each analysed nutrient was calculated as

$$Removal rate={\displaystyle \frac{{Nut}_{beginning}-{Nut}_{end}}{∆t\times A}},$$

(4)

where Nut_beginning is the nutrient mass (g) present in the RW at the beginning of the cycle, Nut_end is the nutrient mass (g) present in the RW at the end of the cycle, Δt is the duration of the cycle (days) and A is the photosynthetic area of the RW (m²).

The overall nutrient removal percentage on the basis of the added effluent was calculated as

$$Removal\%=\left(1-{\displaystyle \frac{{\sum }_1^n\left({Nut}_{end}\times V_{removed}\right)}{{\sum }_1^n\left({Nut}_{effluent}\times V_{added}\right)}}\right)\times 100,$$

(5)

where n is the number of renewal cycles performed for each HRT condition, Nut_end is the nutrient concentration (mg·L⁻¹) present in the RW at the end of cycle n, Nut_effluent is the nutrient concentration (mg·L⁻¹) present in the effluent added for cycle n, V_removed (L) is the culture volume removed after cycle n, and V_added (L) is the volume of effluent added for cycle n. Thus, one overall %Removal value is calculated for each condition tested.

The nitrogen mass balance was calculated through

$$N_{Diss beginning}+N_{Biom beginning}=N_{Diss end}+N_{Biom end}+N_{Other},$$

(6)

where N_{Diss beginning} and N_{Diss end} are the TN masses (g) present in dissolved form in the RW at the beginning and at the end of the cycle, respectively. N_{Biom beginning} and N_{Biom end} are the nitrogen masses (g) present in the biomass from the RW at the beginning and at the end of the cycle, respectively. N_other (g) is calculated from the mass balance, representing other potential routes for N removal.

The phosphorus mass balance was calculated through

$$P_{Diss beginning}+P_{Biom beginning}=P_{Diss end}+P_{Biom end}+P_{Other},$$

(7)

where P_{Diss beginning} and P_{Diss end} are the TP masses (g) present in dissolved form in the RW at the beginning and at the end of the cycle, respectively. P_{Biom beginning} and P_{Biom end} are the phosphorus masses (g) present in the biomass from the RW in the beginning and at the end of the cycle, respectively. P_Other is calculated from the mass balance, representing other potential routes for P removal.

The carbon mass balance was calculated through

$$C_{Diss beginning}+C_{Biom beginning}=C_{Diss end}+C_{Biom end}+C_{Other}$$

(8)

where C_{Diss beginning} and C_{Diss end} are the TC masses (g) present in dissolved form in the RW at the beginning and at the end of the cycle, respectively. C_{Biom beginning} and C_{Biom end} are the carbon content (g) of the biomass from the RW in the beginning and at the end of the cycle, respectively. C_other is calculated from the mass balance, representing other potential routes for C removal.

2.5. Statistical Analysis

For the DW measurements, only one replicate was performed up to 1 March 2024, and thereafter 2 replicates were performed with a maximum standard deviation of 0.01 g·L⁻¹. For N-NO₃⁻ a minimum of 3 replicates were analysed with a maximum standard deviation of 0.2 mg N·L⁻¹. For N-NH₄⁺ a minimum of 3 replicates were analysed with a maximum standard deviation of 0.3 mg N·L⁻¹. For P-PO₄³⁻ a minimum of 3 replicates were analysed with a maximum standard deviation of 0.3 mg P·L⁻¹. For TP, two replicates were performed for the digestion process, and a minimum of 5 replicates in total were analysed with a maximum standard deviation of 0.3 mg P·L⁻¹. For the P content in the biomass, two replicates were performed for the digestion process, and a minimum of 5 replicates in total were analysed with a maximum deviation between replicates of 0.5% (occasionally, only one digestion replicate was performed, and a minimum of 3 replicates were analysed with a maximum deviation between replicates of 0.02%). For the N and C contents in the biomass, a minimum of 2 replicates were performed with a maximum standard deviation of 0.9% and 4.35%, respectively. For TIC, TOC, TC, and TN, 3 replicates were analysed with a maximum deviation between replicates of 2%.

Results for each operational HRT condition are presented in box-and-whisker plot format, where the box comprises the values between the first (25%) and third (75%) quartiles, with a line at the median (50%). The whiskers extend to 1.5 times the interquartile range, and the points past the whiskers are considered outliers.

For the determination of statistical significance of the data the Welch’s ANOVA method was used, and p < 0.05 was the criterion for statistical significance. The Welch’s method was used instead of one-way ANOVA due to the different number of samples in each group and consequent non-compliance with the homogeneity of variance criteria of ANOVA. The results (tables and graphs with error bars) are expressed as mean values of the data obtained for each cycle ± standard deviation.

3. Results

3.1. Environmental Parameters

The environmental parameters are tremendously relevant for the growth of microalgae and, consequently, to the efficiency of the algae-based wastewater treatment, particularly because the systems used in this work are in outdoor conditions. The evolution of solar irradiance, temperature (T) and precipitation (rainfall) throughout the experiment is presented in Figure 2.

There is a significant increase in the average solar radiation per day between the beginning of the experiment (beginning of winter) and the end (end of winter), as expected. This factor is important when comparing the biomass growth throughout the season, since the solar radiation can be a limiting factor for microalgae growth. There was also an increase in the daily temperature, although less substantial.

Periods of intense precipitation occurred within the experiment. Precipitation introduces a dilution effect in the culture, which can have a positive effect since the pollutant concentrations are reduced, and the discharge limits are more easily met. However, this also represents fewer nutrients for microalgae uptake and consequently slower biomass growth, with a higher chance of washout.

3.2. Evolution of Online Measured Parameters Along the Treatment Cycle

Significant changes were observed in the culture’s physico-chemical parameters, as measured in RW2 with in situ probes, caused by the day and night cycle. Due to equipment limitations, measurements of these parameters could only be performed in one RW. RW2 was chosen for this since it operated at HRT values of 5.5 and 4 days, thus covering more variable operational conditions. In addition to changes in air temperature, during the day photosynthesis is occurring, leading to O₂ production and CO₂ consumption. Independently of the HRT employed and operation date range, the evolution of these parameters throughout the day always presented the same pattern, albeit with variations in absolute values. To illustrate this, the evolution of parameter values along a typical cycle is presented in Figure 3.

At the beginning of each cycle, i.e., after the renewal with secondary effluent, the pH of the culture is close to 7, reflecting the prevailing range of effluent pH values. During the day the photosynthesis process promotes an increase in pH, through dissolved CO₂ consumption, which tends to slightly decrease during the night. However, the pH evolves towards stabilisation at values above 10, making it likely that, when present, ammonia nitrogen is partly removed through volatilisation instead of microalgae intake [27]. For HRT at 5.5 days, the average pH attained was 9.2 ± 0.6, and for 4 days, 9.1 ± 0.9 (no in situ pH measurements are available for HRT at 7 days).

Very high dissolved oxygen (DO) values are reached during the day, above 17 mg O₂·L⁻¹ and during the night the DO reaches a minimum value of 7 mg O₂·L⁻¹. The average DO values obtained for the 5.5- and 4-day HRT conditions were 13.1 ± 0.5 (n = 8) and 12.6 ± 0.7 mg O₂·L⁻¹ (n = 15), respectively (no DO values are available for the 7-day HRT condition). These values are actually above the solubility of oxygen in pure water under air at 1 atm pressure (11.5–9.5 mg O₂·L⁻¹ at 10–20 °C, respectively) [32] and reflect the intensity of photosynthesis by the microalgae [27]. The water temperature also suffered marked fluctuations during each cycle, as expected from the air temperature variations between day and nighttime. Minimum, average, and maximum temperature values of 9.5 ± 2.4, 14.4 ± 2.4, and 20.3 ± 3.3 °C (n = 23) were reached, respectively, with no notable differences between the 5.5- and 4-day HRT conditions. Water temperature being mainly correlated to the environmental conditions, namely air temperature, wind, irradiance, and precipitation, although no measurements were performed in RW1, it is likely that the water temperature was very similar between the two RWs, since they were operated simultaneously and side-by-side, with the same water volume and exposed surface area.

3.3. Biomass Growth and Microalgae Consortium

The biomass growth was calculated as an areal productivity using Equation (3), and the individual cycle values for each HRT condition are presented in Figure 4. The final DW concentration values reached for each condition were 0.16 ± 0.07 (n = 30), 0.17 ± 0.04 (n = 14), and 0.10 ± 0.05 g·L⁻¹ (n = 15), for HRT at 7, 5.5, and 4 days, respectively.

Similar biomass areal productivities were achieved for all conditions tested (p = 0.770, F_2,25.6 = 6.12). The cycle values for HRT at 7, 5.5 and 4 days were 3.55 ± 1.55 (n = 30), 3.98 ± 1.85 (n = 14), and 3.69 ± 2.14 g·m⁻²·day⁻¹ (n = 15), respectively. The highest variability between cycle productivity was observed for the HRT at 4 days. Since the fraction of culture volume renewed with effluent at each cycle was the highest in this condition, biomass growth could have been most influenced by the variation in wastewater characteristics. Also, the initial cycle DW values were the lowest, which could have caused a lag phase in the growth profile. Additionally, the 4-day HRT condition was repeated in the beginning and at the end of the winter season, so the changes in environmental parameters (see Figure 2) could have contributed to the higher variability in biomass growth. However, it should be noted that the 7-day HRT condition did not present such high variability even though its cycles were performed throughout the whole season.

A semi-quantitative characterisation of the microalgae consortium was performed through optical microscopy observations, and the relative abundance of each microalgae type, as well as those of the contaminants, is presented in Figure 5.

The microalgae consortium presented little change throughout the season, and the variations in HRT did not appear to affect the consortium composition. The main microalgae genus present in the consortium was Scenedesmus sp., together with small proportions of other genera from the Chlorophyceae class (Tetradesmus, Monoraphidium, and an unidentified genus), from the Cyanophyceae class (Oscillatoria and Pseudanabaena), and from the Bacillariophyceae class (Eunotia and Nitzschia).

More contaminants were typically present in the control RW1 than in the tests in RW2 (Figure 5c,d, respectively). The presence of contaminants can lead to microalgae culture collapse. A lower HRT can reduce the impact of contaminants such as ciliates, rotifers, and amoebas through the effect of renewing RW contents before the contaminants reach abundance levels that can affect the culture. On the other hand, rainfall periods (see Figure 2) often coincided with higher observed abundances of contaminants, suggesting that contaminants may have been carried by rainwater or affected through other unknown mechanisms.

3.4. Nutrient Removal

The dissolved nitrogen fractions N-NO₃⁻, N-NH₄⁺, and TN were monitored in the secondary effluent and in the microalgae cultures in the beginning and at the end of each cycle, and the calculated areal removal rates (Equation (4)) for each are presented in Figure 6, as well as the dissolved TN concentration values at the end of each cycle.

Ammonia nitrogen was almost always completely removed, reaching concentration values below the quantification limit of the analytical procedure employed (<0.78 mg N·L⁻¹), irrespective of the cycle’s initial concentration. Removal rate values for N-NH₄⁺ were mainly related to the presence of this nutrient in the wastewater, and the effect of HRT was not significant (p = 0.105, F_2,23.3 = 2.48). The N-NO₃⁻ levels reached for HRT values of 7, 5.5 and 4 days were <0.24 mg N·L⁻¹, <0.27 mg N.L⁻¹, and <1.79 mg N·L⁻¹, respectively. There was a significant impact of HRT on the N-NO₃⁻ removal rate (p = 0.007, F_2,22.8 = 6.12), i.e., the longer the HRT, the lower the rate.

In terms of total nitrogen, the final levels achieved always complied with the discharge limit under consideration (8 mg N·L⁻¹), and there were no significant differences between the operation conditions tested (p = 0.387, F_2,19.1 = 0.999). However, TN removal rate values show significant differences between the tested HRT conditions (p = 0.0001, F_2,25.6 = 12.8). For the TN removal rate, the lower the HRT, the higher the rate. This is expected since, for the same cycle duration, similar final concentration values were reached for all HRT values, but from higher initial concentrations for the lower HRT operations. A lower HRT implies that a larger volumetric proportion of secondary effluent is fed at the beginning of each fill-and-draw operational cycle, carrying a larger load of nutrients, which results in the aforementioned higher initial nutrient concentrations. The 4-day HRT condition sometimes resulted in final concentration levels slightly higher than those of the other conditions, resulting from higher N loading rate values (Figure 6d). The overall values for TN removal percentage from the secondary effluent were 91.3%, 92.7%, and 86.8% for the 7-, 5.5-, and 4-day HRT conditions, respectively.

The dissolved phosphorus fractions P-PO₄³⁻ and TP were also monitored in the secondary effluent and in the microalgae cultures in the beginning and at the end of each cycle. The dissolved concentration values achieved at the end of each cycle, as well as the calculated areal removal rate (Equation (4)), are presented in Figure 7.

As observed in the results for nitrogen, the P-PO₄³⁻ (p = 0.136, F_2,28.4 = 2.14) and TP (p = 0.216, F_2,27.9 = 1.62) removal rates were higher for the lower HRT values, although not significantly. The final TP levels achieved for HRT at 7, 5.5, and 4 days were <0.63 mg P·L⁻¹, <0.62 mg P·L⁻¹, and <1.00 mg P·L⁻¹, respectively, excluding an outlier value caused by a spike in the secondary effluent. The 7-day and 5.5-day HRT conditions almost always allowed the final TP concentration to reach the objective of 0.5 mg P·L⁻¹, while the 4-day HRT failed to do this in more than 25% of the cycles performed (p = 0.037, F_2,21.7 = 3.85). The overall values for TP removal percentage from the secondary effluent were 91.3%, 90.8%, and 81.5% for the 7-, 5.5-, and 4-day HRT tests, respectively.

The dissolved carbon fractions TC, TIC, and TOC were also measured in the secondary effluent and in the microalgae cultures. The calculated areal removal rate (Equation (4)) values for each cycle are presented in Figure 8.

In terms of the dissolved carbon fractions, the measured removal was mainly of inorganic carbon, while the organic carbon levels in the cultures presented negligible changes from the beginning to the end of each cycle (Figure 8a). Therefore, the TIC and the TC removal rate values are very similar (Figure 8b,c). There are significant differences between the three conditions tested in terms of carbon removal rates (TIC: p = 0.016, F_2,25.6 = 4.92; TC: p = 0.011, F_2,25.8 = 5.42). The 4-day HRT condition shows a higher variability in the removal rates than the other conditions. The 5.5-day HRT condition shows higher carbon removal rate than the 7-day condition. In terms of overall TC removal percentage from the secondary effluent, values of 46.0%, 56.2%, and 27.1% were achieved for the 7-, 5.5- and 4-day HRT conditions, respectively.

3.5. Mass Balances on N, P, and C

Mass balances were performed for N, P, and C, using Equations (6)–(8), respectively. From the calculated inputs to each cycle (left-hand side of the equations) and the calculated outputs from the suspended biomass and from the fractions dissolved in the liquid (right-hand side of the equations), the N_Other, P_Other, and C_Other outputs were computed. The ratio between each of the three nutrient outputs and the total input/output is presented in Figure 9.

For the 5.5-day HRT condition, the fraction of the total N, P, and C remaining dissolved in the water was lower than for the other conditions. The fraction of the total N, P, and C recovered in the biomass decreased with decreasing HRT, i.e., the higher the HRT, the higher the percentage of the nutrient recovered in the biomass and not lost through other removal mechanisms. The N_Other portion increased with decreasing HRT values for all nutrients. A positive N_Other, P_Other, or C_Other value indicates there is an exit (removal) pathway from the system that is not being measured. For nitrogen and carbon, this loss is most likely an emission to the atmosphere; therefore, no recovery is possible. For phosphorus, the most probable exit pathway is precipitation as insoluble salts, and the nutrient can still be recovered through biomass harvesting. It should be noted that elemental analysis was performed here on biomass samples after washing with distilled water. This procedure could have washed off at least part of any precipitated P salts associated with the biomass concentrate.

4. Discussion

Table 3. Overview of the results achieved in this study and in other pilot-scale studies with RW-type reactors treating secondary wastewater with very low residual pollutant concentrations, reported in the literature. DOC—dissolved organic carbon; IC—inorganic carbon; DTN—dissolved total nitrogen; DTP—dissolved total phosphorus.

Operation Conditions	Influent Pollutant Levels (mg·L−1)	Pollutant Removal (%)	Outlet Pollutant Levels (mg·L−1)	Biomass Final Level and Productivity (Dry Weight)	Ref
Continuous Duration: 110 days HRT: 10 days pH: 8–9.5 Volume: 530 L Area: 1.93 m2	TN: 24.9–26.2 TP: 1.8–2.2	TN: 56–77 TP: 51–64	TN: 14 2–20 2 TP: 0.9 2–1.4 2	0.16–0.25 g·L−1 Maximum: 8.26 g·m−2·d−1	[14]
Continuous Duration: 40 days HRT: 4 days Artificial illumination (>400 µmol m−2s−1) Volume: 8 m−3 Area: 18 m2	COD: 12.8 ± 1.9 TN: 7.7 ± 0.4 N-NO3−: 3.7 ± 0.5 N-NH3: 0.2 ± 0.1 N-NO2−: 0.04 ± 0.02 TP: 1.1 ± 0.5 P-PO43−: 0.8 ± 0.2	TN: 30 1–75.5 TP: 20 1–86.0	TN: 2 1–7 1 TP: 0.1 1–0.6 1	0.67–1.7 g·L−1 11.7–21.3 g·m−2·d−1	[15]
Continuous Duration: 193 days HRT: 4 days Volume: 380 L Area: 1.52 m2	TN: 7.89–22.2 DTN: 7.6–21.5 2 N-NH4+: 0–12 1 N-NO2−: 0–1.5 1 N-NO3−: 5–20 1 TP: 0.45–3.41 DTP: 0.4–3.2 2 P-PO43−: 0.5–4 1 TOC: 4.70–10.7 DOC: 3.9–8.9 2 IC: 5–35 1	DTN: 70–100 2 DTP: 60–100 2	DTN: 0–15 1 DTP: 0–2 1 DOC: 5–12 1 IC: 13–35 1	0.025–0.25 g·L−1 7.63–10.4 g·m−2·d−1	[16]
Semi-continuous Duration: 89 days HRT: 7 days Volume: 700 L Area: 5 m2	N-NO3−: 5.2 ± 1.8 N-NH4+: 5.0 ± 4.1 TN: 9.4 ± 4.1 P-PO43−: 2.1 ± 1.3 TP: 2.3 ± 1.3 TIC: 22.0 ± 6.9 TOC: 6.6 ± 1.8 TC: 28.5 ± 8.4	TN: 91.3 TP: 91.3 TC: 46.0	TN: 0.5–1.2 TP: 0.2–0.8 TC: 9.0–20.9	0.06–0.31 g·L−1 0.6–7.6 g·m−2·d−1	This study
Semi-continuous Duration: 40 days HRT: 5.5 days Volume: 700 L Area: 5 m2		TN: 92.7 TP: 90.8 TC: 56.2	TN: 0.5–1.1 TP: 0.2–0.6 TC: 9.0–22.6	0.11–0.22 g·L−1 0.6–7.8 g·m−2·d−1
Semi-continuous Duration: 35 days HRT: 4 days Volume: 700 L Area: 5 m2		TN: 86.8 TP: 81.5 TC: 27.1	TN: 0.5–2.7 TP: 0.2–1.0 TC: 8.9–31.3	0.03–0.16 g·L−1 1.1–9.3 g·m−2·d−1

¹ Values presented in graphs. ² Values obtained through calculations on presented data.

summarises the results of the current work and compares them with data found in the literature for similar studies. It should be highlighted that there is not much information regarding the use of open, RW-type microalgae systems for secondary municipal wastewater treatment (i.e., tertiary treatment) at pilot scale. Most reports are either from laboratory-scale experiments or typically focus on primary effluent treatment when performed at larger scales.

The study by Arbib et al. [14] focused primarily on the comparison between open and closed photobioreactors for enhancing biomass production in tertiary wastewater treatment. For this, a tubular system and an RW were used, only the latter being the focus of this discussion. The system was first inoculated with Scenedesmus obliquus, and some HRT optimisation was performed in batch mode. An HRT value of 10 days was selected, and the system was operated in this condition, in a continuous regimen, for 110 days, with fresh wastewater feeding only performed during the light (day) period. The treatment efficiency was evaluated through TN, TP and COD measurements.

The study by Min et al. [15] focused on tertiary treatment using an RW system installed in a greenhouse, with radiation supplementation using an underwater illumination device coupled to an LED lamp and a light collector. Hydrodictyon reticulatum was used as inoculum, and the system was operated for 40 days in a continuous regimen with an HRT of 4 days. The treatment efficiency was evaluated through TN and TP measurements.

The study by Takabe et al. [16] used secondary effluent for the growth of indigenous microalgae species, assessing the effects of different HRT values and weather conditions. For this, three reactors were used, one with a volume of 380 L and the other two of 15 L, operated in a continuous regimen at 4-, 2- and 7-day HRT values, respectively. For this discussion we will focus on the results achieved for the pilot-scale reactor, an open pond system completely mixed with mechanical stirring. This reactor was operated for 193 days at the 4-day HRT value, and pH control was implemented through CO₂ addition to keep the culture pH within the 7.7–8.0 range. The treatment efficiency was evaluated through TN, dissolved TN, ammonia nitrogen, nitrate, nitrite, TP, dissolved TP, phosphate, TOC, dissolved organic carbon, and inorganic carbon measurements.

In the present study, a natural, indigenous microalgae consortium was developed, like in the Takabe et al. study [16], instead of inoculating a specific species [14,15]. The same microalga (Scenedesmus sp.) dominated the consortium throughout the operation period, enduring the harshest season conditions, e.g., rainfall and low temperature and radiation levels (Figure 2), as well as the unfavourable pH and DO values attained (Figure 3), known to impair algal productivity [27]. Takabe et al. [16] went further into microalgae identification, and their natural consortium suffered more marked changes throughout the experiment than in the present study. However, Scenedesmus species also dominated the culture very often during the study, together with some periods when microalgae either from the Bacillariophyceae or from the Chlorophyceae classes showed higher abundances.

The final biomass DW values reached in our study were lower for an HRT of 4 days than for the other two conditions, presenting similar values to those reported by Arbib et al. [14] and Takabe et al. [16]. Min et al. [15] obtained higher DW values, possibly because of the higher radiation levels achieved with their underwater illumination device, complementing the sunlight input. In terms of productivity, the results achieved in our study were marked by high variability within each condition tested when compared to those of the other three reports in Table 3, with maximum values typically lower than theirs. Operation in continuous mode, leading to less variable and less extreme pH values, together with pH control in one case [16], may have been the main reason for the higher biomass productivity values when compared to those obtained under the semi-continuous regimen of the present study. In terms of areal productivity, lower values were typically reached when compared to those reported by Arbib et al. [14], Min et al. [15] and Takabe et al. [16]. This may be related to the lower nitrogen concentration present in the influent [14,16] and to irradiance limitation [15].

In terms of treated water quality, the TN levels in the outlet were much lower than the discharge limit imposed by the new UWW directive (8 mg N·L⁻¹) for all HRT conditions tested. When compared to the other studies in Table 3, the values obtained were always lower in our study, although it should be considered that the TN contents in the feed wastewater were higher in two of the studies [14,16], possibly also contributing to their higher biomass productivity (Table 3). The outlet TP levels obtained for HRT values of 7 and 5.5 days almost always complied with the new discharge limit (0.5 g P·L⁻¹) and were similar to those obtained by Min et al. [15], at an HRT of 4 days. In the present study, the 4-day HRT condition led to more than 25% of the outlet TP values being above the 0.5 g·L⁻¹ target. These results were similar to those reported by Arbib et al. [14] and Takabe et al. [16] when operating at HRT values of 10 and 4 days, respectively.

Microalgae cultivation studies on wastewater which disclose mass balances to the nutrients are very scarce [5], impairing the tracking of whether the nutrients are being recovered or lost, e.g., to the atmosphere. In the present study mass balances to the N, P, and C loads in the system were performed, showing that the fates of the nutrient inputs varied, depending on the implemented HRT value.

When considering the balance in nitrogen in Equation (6), if the N_other fraction reveals a positive value, then nitrogen was most likely released to the atmosphere through volatilisation, due to the high pH levels attained (ammonia stripping is probable at pH values above 8 [27]. If the value results in negative, then additional sources of nitrogen were brought into the system. For instance, nitrogen could be captured from the atmosphere by nitrogen-fixing bacteria such as some cyanobacteria [33] or introduced by bird droppings. Concerning the balance to phosphorus (Equation (7)), if the P_Other fraction reveals a positive value, phosphorous most likely is removed by precipitation of its insoluble salts, again due to high pH levels [27]. It should be remembered that biomass samples were washed with distilled water before being subjected to elemental analysis (see Section 2.3). Thus, most of the precipitated phosphate would not be quantified as P_Biom, although in typical harvesting processes this precipitated phosphorous fraction could be retrieved with the biomass. If the P_Other value is negative, sources of phosphorous other than the feed wastewater must be considered, e.g., bird droppings. Also, the mass balance assumes that, in the cycle renewal process with new wastewater, all the insoluble phosphate remaining from the previous cycle will be dissolved. However, if the volume added is not sufficient for this to occur, the P_Other value can also result in negative since the precipitated phosphorus may not be quantified in the liquid at the beginning of the cycle. Finally, in the balance of carbon of Equation (8), if the C_other fraction has a positive value, carbon could have been released to the atmosphere, though the high pH values could also be involved in carbonate precipitation [34]. If the value is negative, then CO₂ was most likely captured from the atmosphere and used by the microalgae in the photosynthetic process.

As can be observed in Figure 9, higher HRT values led to higher removal of all three nutrients through biomass uptake, while the undetermined fraction (N_Other, P_Other, and C_Other) increased with decreasing HRT values. Removal through biomass uptake is ideal, namely for nitrogen since the other outlet routes (treated water or atmosphere) necessarily involve a negative environmental impact. In the present study, volatilisation apparently could not be avoided, representing up to 20% of the nitrogen input. So, even though the TN consumption rate grew with decreasing HRT values (Figure 6c), the lower HRT condition may not be the best for effective nitrogen recovery. Since loss of nitrogen to the atmosphere likely occurs in the form of ammonia, the occurrence of extensive nitrification in the secondary WWT stage would be a positive aspect, i.e., avoiding high N-NH₄⁺ values such as those registered during this experimental period (see Table 1). In the case of phosphorus, the P_Other fraction is most likely inorganic phosphorus precipitation, and this fraction can be easily recovered with the biomass and valorised. So, the higher TP consumption rates observed for lower HRT values (Figure 7b) can effectively represent an increased P recovery potential, though more in inorganic form (up to 20% of the input load, Figure 9b). Examining the values of the C_Other fraction in Figure 9c, negative values prevail in the 7-day HRT condition, corresponding to CO₂ uptake from the atmosphere, which is a step towards the reduction in the carbon footprint. Moreover, the dominant fraction is removed in the biomass, entailing the possibility of organic carbon valorisation, e.g., through biomethanation. Since the carbon input to the system is largely in the form of inorganic carbon (Table 1) and the high pH values attained in the RW culture, the hypothesis of positive C_Other values (Figure 9c) being due to carbonate precipitation should be considered. Thus, the recovery and analysis of inorganic precipitates originating in the RW operation cycle is a desirable feature in further studies so as to allow a more detailed mass balance and the effective assessment of the carbon footprint of the process.

5. Conclusions

Tertiary wastewater treatment using an open microalgae system culturing an indigenous consortium was assessed, under different HRT values, throughout the winter season. The 7- and 5.5-day HRT conditions mostly allowed the treated wastewater to comply with the stricter discharge limits required by the new UWW directive. The 4-day HRT condition did not reach compliance for phosphorus levels approximately 25% of the time; therefore, it is not ideal for this season. Lower HRT values resulted in higher removal rates, although there were no significant changes in biomass productivity. The different HRT conditions seem to lead to different nutrient removal mechanisms, with the higher HRT conditions being associated with higher recovery of nutrients in the biomass. This aspect should be considered when optimising the process, namely in terms of its environmental impact and circularity. This technology shows great promise but must be further studied at pilot scale in field conditions to be implemented at larger scales. The performance variability it exhibited calls for experimentation throughout all the yearly seasons, for the impact of atmospheric conditions to be evaluated and incorporated in system design and operational control. More detailed mass balances should be performed in future studies, namely identifying and quantifying the inorganic precipitates involved, to effectively assess the environmental impact of the process. Studies on biomass harvesting and precipitate recovery are also required, since they are decisive for the feasibility of implementing a large-scale microalgae system for tertiary wastewater treatment.