Steady state modeling of autotrophic membrane bioreactor – a new approach to quantify biomass

2.1 Experimental set up

The experimental setup consisted of a 30-L of aerobic bioreactor equipped with a continuous pH controller and a 0.8 L submerged hollow fiber membrane module (0.05 µm pore size and 0.2 m² of surface area) (Figure 1). Due to the high mixing rate, the reactor and the membrane module were considered perfectly mixed. The concentrated synthetic feed solution, the diluting water and the permeate were injected or extracted by peristaltic pumps. Aeration was continuously provided through membrane diffusers at the bottom of the reactor and just below the fibers in the membrane module enabling to operate without dissolved oxygen (DO) limitation.

2.2 Biological conditions

Two successive experiments were carried out under the operational requirements, as shown in Table 1. At the beginning of the first run, the reactor was filled with sludge inoculums from a domestic wastewater plant operated with low organic loading rate (<0.1 kg COD/kgVSS/d). The reactor was then fed with a synthetic solution containing ammonium chloride (NH₄Cl) with additional phosphorus salts as diammonium phosphate (NH₄)₂HPO₄. Sodium carbonate (Na₂CO₃) was added to ensure the necessary alkalinity for the nitrification reaction. No organic carbon was in the reactor, as the feeding solution's COD/N ratio was always zero. Other elements (Mg²⁺, K⁺, etc.) were supplied by tap water used as diluent.

During the first period, a nitrogen load rate (NLR) of 0.22 kgN/m³/d and a sludge age of 20 days were imposed. For the second run, the NLR decrease, from 0.44 to 0.374 kgN/m³/d, and the SRT was set at 40 days. At the beginning of the second run, sludge extraction was temporarily halted to achieve the expected concentration values of total suspended solids rapidely. The bioreactor was operating for 125 successive days. The monitoring during run I and II was done for 46 and 79 days, respectively.

During these runs, the pH was adjusted in the range of 7.5 ± 0.5 by the ez-control system which an automatic pH controller using a conventional proportional integral derivative (PID) control. The executive element was a peristaltic pump dosing the acid solution when the pH is increasing and alkaline solution when the pH is decreasing.

2.3 Analytical methods

Total suspended solids (TSS) and volatile suspended solids (VSS) were analyzed according to the Standard Methods (APHA, 2005). Ammonia, nitrate, and nitrite in the influent and effluent were measured using a colorimetric method (HACH DR/2500). The analysis of extracellular polymeric substances (EPS), which are polymer materials secreted by cells, was accomplished through the determination of protein and carbohydrate content according to Frolund, Griebe, and Nielsen (1995) and Dubois et al. (1956), respectively.

2.4 Respirometric analysis

A respirometric measurements was set up to study the kinetics of biological reactions by monitoring the evolution of DO concentration in the reactor throughout time. There are several methods for measuring the respirometry needs of a bacterial population (Gasmi, Heran, Hannachi, & Grasmick, 2015). In this work, we relied on one method that was carried out in a closed batch reactor, because it has the advantage to overcome the oxygen transfer phenomena from air to the environment.

The following protocol was adopted to perform these measurements: A volume of 250 ml of sludge from the continuous reactor is taken and placed in another batch and stirred reactor. The pH and temperature were controlled to be not a limiting factor to the biological reaction. The requirement for oxygen is evaluated by measuring the instantaneous concentration of DO in the medium using an oximeter (Oxi 330i). The rate of DO consumption over time is known as Oxygen Uptake Rate (OUR (mgO₂/L/d)). The experimental device used is presented in Figure 2.

The respirometric tests were carried out in endogenous respiration. The sludge in bioreactor was aerated without supplying of substrate for 24 hours; thus, it can then be assumed that the biodegradable substrates were consumed during this time. This duration is sufficient to achieve a constant total endogenous OUR noted OUR_endt. A sample from bioreactor was then transferred to the batch reactor to monitor the DO over time. Moreover, two specific inhibitors were added to the sludge sample placed in the batch reactor to quantify the relative activity of the different populations present in the bacterial culture. The first is the allythiourea solution (ATU) (20 mmol.L⁻¹) that is known as an inhibitor of autotrophic microorganisms and, more particularly, of the Nitrosomonas bacteria (Gorska, Gernaey, Demvunck, Vanrolleghem, & Verstraete, 1995). The second inhibitor is the sodium azide (24µM) or of the sodium chlorate ClO_3⁻ (2.3 mol/L), known as The Nitrobacter inhibitor (Chandran and Smets, 2000).

The nitrification reaction is the net result of two distinct processes (Heil, Vereecken, & Brüggemann, 2016).

• -

  Oxidation of ammonium (NH₄⁺) to nitrite (NO₂⁻) by nitrosomonas bacteria:

• -

  Oxidation of nitrite (NO₂⁻) to nitrate (NO₃⁻) by the Nitrobacter bacteria:

Figure 3 represents an example of a curve obtained after successive additions of both inhibitors. In fact, after reaching the endogenous respiration, a sample was taken from the membrane bioreactor for DO monitoring as it aformentioned. The first slope of the line segment AB represents the total oxygen uptake rate (OUR_endt). After the injection of ATU, the Nitrosomonas bacteria will be inhibited. The second slop of segment BC represents the oxygen uptake rate resulted only from the endogenous respiration of Nitrobacter and heterotrophic bacteria. The injection of sodium chlorate inhibits the Nitrobacter bacteria, therefore, the slope of segment CD represents the endogenous oxygen uptake rate only of heterotrophic bacteria. The differences in slope obtained allow going back to the specific oxygen requirements of the different species of the sample.

The slope of the linear portion of the DO profile with time is the OUR.

3.1 Performance of membrane bioreactor (MBR)

Table 2 summarizes the effluent qualities under steady-state condition. During the first run the TSS and VSS concentrations decrease compared to the effluent, this was likely due to (i) the continuous sludge withdrawn (SRT = 20 days) and (ii) the decrease of heterotrophic bacteria related to the lack of organic substrate in the influent. However, during the second run the TSS and VSS concentration increase due to the increase of SRT (40 days) and stop sludge withdrawal at the beginning of the second run. The average effluent TSS and VSS concentrations were: 485 ± 51 and 351 ± 38 mg/L for run I, 842 ± 65 and 748 ± 46 mg/L for run II.

The monitoring of nitrogen species throughout the study, shows that MBR was able to achieve satisfactory nitrogen removal, nitrate were the major nitrogen species in the effluent of the MBR suggesting complete nitrification in the treatment process. The removal efficiencies were more than 94% and 96 % for run I and II respectively. It should mention that, the monitoring of DO concentration exhibit that there is no oxygen limitation for nitrification reaction with a value of 6 mg/L.

3.2 Effect of SRT on fouling membrane

The major constraint for MBR application was membrane fouling since it cause an increase of operational cost (Rahman et al., 2023). Tracking transmembrane pressure (TMP) or permeate flux variations over time are the two conventional methods for membrane-fouling monitoring. In fact, the flow of wastewater through a porous membrane was described by the Darcy's law given by Equation (3):

• J is the permeate flux (L·m⁻²·h⁻¹).

• TMP: The transmembrane pressure (Pa).

• µ: The viscosity (Pa·s).

• R_t: The total resistant which the sum of The fouling resistance and membrane resistance.

The method of R_t determination has been widely investigated in previous works (Gasmi, Heran, Hannnachi, & Grasmick, 2012, 2013). The monitoring of TMP with time allowing the determination of fouling rate that is defined by the evolution of resistance or TMP with time. Table 3 presents the fouling rate obtained in our study as well as for other works.

The SRT has an important effect on the sludge properties, including the TSS concentration, the presence of extracellular polymeric substances (EPS) and soluble microbial products (SMP) resulted from bacteria activities and known among the responsible of fouling membrane. As it showed in Table 3, the membrane fouling rate in this study was highest at SRT equal to 20 days operation compared to SRT equal to 40 days. These results were consistent with some other studies that suggested MBRs operated under a prolonged SRT tend to have a lower fouling potential (Ouyang & Liu, 2009; Deb et al., 2022). In this study, It seems that the increase of TSS during the second run hasn’t affected the fouling propensity. Thus, higher air flow intensity through membrane (200NL/h) was sufficient to prevent sludge deposition on surface membrane. However, it was found during this study that the difference of EPS concentration inside the bioreactor (EPSs) and in the permeate (EPSp), (EPSs-EPSp) was found equal to 5-20 mg/L for run I and 2-12mg/L for run II. Therefore, the membrane has a significant role in the quality of permeate regarding to the soluble fractions and these materials contribute to fouling mechanism. Ahmed et al. (2007) found that the bound of EPS per biomass unit increase as the SRT decreased. Nevertheless, some other researchers observed that at a long SRT, the SMP and EPS concentrations were higher (Faridizad et al., 2022). Huang et al. (2011) found that in short SRT, the microorganisms metabolized more actively, however less SMP were produced, which restricted biofilm growth and membrane fouling.

3.3 Steady state equation developing

3.4 X_BA and X_BH evaluation by respirometric measurements

The respirometric measurements in endogenous condition was used to calculate the biomass concentration (X_BH and X_BA). The obtained values were compared with those obtained from the equations proposed. Moreover, the quantification was done by integrating the heterotrophic activity developed on biodegradable products resulting from the lysis of autotrophic bacteria.

The steady state was reached in runs I and II. The respirometric measurements are made in endogenous condition without inhibitor (OUR_endt) and after injection of two inhibitors (OUR_endhet).

The concentrations of autotrophic and heterotrophic biomasses can be calculated using the measured values of the OUR_endt, OUR_endhet and OUR_endaut given by Equations (12), (15) and (16).

An approximation of µ_BHend is made based on Equation (25). Finally, the value of soluble biodegradable substrate in the endogenous state is calculated according to Equation (26) (Héran, Wisniewski, Orantes, & Grasmick, 2007).

Since the bioreactor was operated under two SRTs, Ss and µ_BHend have two different values. The obtained values were 0.66 and 0.64 d⁻¹ for SRT equal to 20 and 40 days, respectively. An average of 0.65d⁻¹ value was adopted for µ_BHend. The monitoring was done in the run I and II after reaching the steady state condition during days: 20, 40, 95, 115, and 120. The measurement's results were summarized in Table 7.

Table 8 gives the kinetic and stoichiometric parameters used to calculate X_BA and X_BH with a comparison to the default ASM1 values. One stoichimetric parameter (e.g Y_A) and three kinetic parameters (e.g. b_A, µ_Amax, K_NH) were adjusted from the labscale tests (Gasmi, Heran & Hannachi, (013), so that the predictions of the model accurately agreed with the actual performance of MBR. Once a steady state was reached, the maximum growth rate of nitrifiers obtained in this study is compared to literature value as obtained by Choubert, Racault, Grasmick, Beck, and Heduit (2005). These authors analyzed and simulated the performance of activated sludge bioreactor for the treatment of nitrogen pollution. Since, µ_Am and b_A are very correlated, their simultaneous identification need a stabilized active biomass concentration (i.e. steady state condition). Therefore, there is one unique couple (µ_Am, b_A) that can predict the nitrogen elimination performance in MBR, b_A was set at 0.14 d⁻¹. The values of the autotrophic yield (Y_A) are close to the default values. Regarding to the half-saturation coefficient for ammonia nitrogen (K_NH), the value obtained in this study (1.6 mgN.L−1), Leyva-Díaz, González, Muñío, and Poyatos (2015) have been obtained a close value in the treatment of nitrogen compounds by moving bed biofilm reactor-membrane bioreactor (MBBR-MBR). Moreover, Mannina et al. (2018), found that the factor mostly influencing the total nitrogen removal is the bacteria affinity factor for O₂, confirming the interest of respirometric measurement for biomass quantification.

Table 9 recapitulates the X_BA and X_BH values obtained after respirometric measurements and steady state equations and the deviation percentage.

Under steady condition, measuring OUR under endogeneous conditions allowed the evaluation of autotrophic and heterotrophic biomasses through Equations (11), (14) and (15). At day 40 (SRT = 20 d), the steady state was established and the percentage deviation of active biomass concentration between the equations developed in steady state and those obtained by respirometric analysis were 3.37% and 15 % for X_BA and X_BH, respectively. Whereas for SRT equal to 40 d, the deviation percentage was 13 and 4 % for X_BA and X_BH, respectively. The results demonstrated the effectiveness of respirometric measurements in access to the active biomass in the bioreactor.

Nevertheless, the highest percentage of deviation could be explained by the specific limitations of the respirometric method, which influenced the result’s precision. The sensor's measurement accuracy and response time are severely constrained by the aging of the probe membrane, resulting in low accuracy and poor stability (Nei & Lillenberg, 2009). Bubbles on the sensor’s surface can also generate a signal disturbance and cause imprecision in the concentration measurement.

Although inlet flow is devoid of organic substrate, the heterotrophic biomass exists in the bioreactor, confirming that the heterotrophic bacteria are developed on biodegradable products resulting from the lysis of bacterial autotrophs.

3.5 Simulation of active biomass evolution

To better understand the evolution of biomass concentration in the membrane bioreactor (MBR) over time, the bioreactor has been modeled using GPS-X software. Indeed, GPS-X is a dynamic wastewater treatment plant simulator, which allows the simulation of a variety of different biological wastewater treatment systems like activated sludge systems with reactors functioning under different situations (aerobic, anoxic, anaerobic), including sludge return and internal recirculation streams, batch reactors, and MBR.

Simulation results (Figure 6) show the evolution of the concentration of the different bacterial species over time for two SRT values (20 and 40 days). The simulation was conducted using kinetic and stoichiometric parameters mentioned in Table 7.

Figure 6 shows that the reactor's autotrophic (X_BA) and heterotrophic bacteria (X_BH) concentrations increase over time until reaching steady-state conditions. The concentration X_BA was much higher than the X_BH due to the imposed operating condition COD/N equal to 0 and it confirmed the idea that the heterotrophic biomass growth deponds on the substrate from the autotrophic biomass. Same pattern of biomass evolution as the TSS concentration.

A good agreement between experimental and simulated results was observed. For an SRT equal to 20 days, the average concentrations of X_BA and X_BH obtained from the respirometric measurements were about 212 (considering the value obtained on day 40, in which the steady-state condition was more established) and 65 mgCOD/L, respectively. These concentrations are close to those obtained by simulation (220 and 73 mgCOD/L for X_BA and X_BH, respectively). The same trend was noticed between experimental and simulation on steady-state bacteria concentrations at SRT 40 (e.g. for autotrophic bacteria, the obtained values were 393 (day 120) and 390 mgCOD/L for experimental measurements and simulation, respectively). The increase of SRT from 20 to 40 days, leads to increase the TSS, VSS concentrations as it mentioned in Table 2, as a result increasing of bacterial concentrations. These results revealed that the ASM1 model had been successfully established to simulate the biological process of the membrane bioreactor. Table 10 gives some results of MBR modeling by other researchers. Baek et al. (2009) reported in their research that the simulated results of X_BH evolution in MBR treating dilute municipal wastewater increase by SRT increasing and has the same pattern as the TSS evolution.

The simulation’s results confirmed the importance of respirometric tools for biomass quantification. In addition, the autotrophic bacteria quantity represent approximately 60 ± 6% and 52 ± 3% of the total volatile solids for run I and II, respectively. In fact, this result highlights the importance of autotrophic biomass quantification, as the measurements of apparent removal rates of ammonium (i.e. expressed through the VSS concentration(kgN/kgVSS/h)) seem irrelevant to characterize their specific activity.