Investigation of factors affecting sodium alginate fouling mechanisms in a microfiltration process under non-constant-flux and non-constant-pressure conditions Open Access

A_eff

membrane surface area (m²)

C

solution concentration (kg/m³)

dJ'/dt

initial normalized permeate flux attenuation rate (s⁻¹)

J

permeate flux (L/(m² h))

J₀

permeate flux of pure water before microfiltration (L/(m² h))

J₁₂₀

permeate flux at 120 min of microfiltration (L/(m² h))

J_f

final permeate flux of pure water after the cake layer was removed (L/(m² h))

J'

normalized permeate flux (s⁻¹)

normalized permeate flux at the beginning of the initial 10 min microfiltration time (s⁻¹)

normalized permeate flux at the end of the initial 10 min microfiltration time (s⁻¹)

K_a

ratio of the blocked membrane surface to the filtrate volume (m⁻¹)

M_SA

amount of SA deposited on the membrane surface (kg/m²)

M_SA-PAC

amount of SA and PAC deposited on the membrane surface (kg/m²)

R_c

cake resistance (m⁻¹)

R_f

sum of R_m and R_p (m⁻¹)

R_m

clean membrane resistance (m⁻¹)

R_p

pore-blocking resistance (m⁻¹)

R_t

total resistance (m⁻¹)

R²

correlation coefficient

t

filtration time (s)

TMP

transmembrane pressure (Pa)

TMP₀

initial transmembrane pressure of pure water filtrated through the clean membrane (Pa)

TMP₁₂₀

TMP at 120 min of microfiltration (Pa)

TMP_f

final transmembrane pressure of pure water after the cake layer was removed (Pa)

TMP'

normalized transmembrane pressure

V

cumulated volume of permeate (m³)

Greek symbol

ɑ

specific cake resistance (m/kg)

ρ

density (kg/m³)

μ

viscosity of the permeate (Pa s)

Membrane separation technology has been widely used in wastewater treatment and reuse due to its high effluent quality and operational flexibility. Filtration in membrane systems is often categorized as either constant-flux or constant-pressure. In constant-pressure filtration, the filtrate is driven through the membrane pores via the driving force provided by compressed nitrogen (Meng et al. 2019), with decreasing permeate flux with the filtration time due to the development of membrane fouling. In constant-flux filtration, the driving force that drives the filtrate through the membrane pores can be generated by vacuum pumps, peristaltic pumps, or diaphragm pumps (Du et al. 2017; Zhao et al. 2017; Kirschner et al. 2019; Xiong et al. 2019), and the transmembrane pressure (TMP) will increase to maintain the set flux (Sreedhar et al. 2022).

Several simultaneous variations in permeate flux and TMP have been reported. Banu et al. (2008) found that an increase in TMP caused a slight decrease in flux from 17 to 16.8 L/(m²·h) over a period of anaerobic-anoxic-oxic membrane bioreactor (A2O-MBR) operation. Li et al. (2019) observed a simultaneous decrease in permeate flux and an increase in TMP with time over 30 days of A2O-MBR operation. Zhao et al. (2017) reported that the permeate flux decreased slightly from 14 to 9 L/(m²·h) during each filtration cycle, which was coupled with increases in TMP during long-term testing in an oxygen-limiting MBR. Research related to the simultaneous changes in TMP and J in the membrane system has been scarce. Therefore, it is necessary to explore the law of simultaneous changes in TMP and J to better understand the fouling mechanism in microfiltration.

The fouling mechanism can be attributed to inorganic, organic, and microbial pollutants, which were deemed to be responsible for cake formation and pore-blocking (Le-Clech et al. 2006). Polysaccharides, one of the main components of extracellular polymeric substances (EPS) utilized by microorganisms, will adhere to the membrane surface, making it prone to adsorbing suspended microorganisms and colloids, causing serious membrane biofouling problems in the MBR (Tansel et al. 2006; Liu & Sun 2010; Lin et al. 2014). Membrane fouling will also increase the filtration resistance of the membrane system, reducing the water production rate and increasing water production costs (Tian et al. 2013; Du et al. 2017). Therefore, the mitigation of membrane fouling is an important measure to effectively improve membrane system efficiency and reduce operating costs. As a natural biological polysaccharide, sodium alginate (SA) has often been used by researchers as a model polysaccharide alone or mixed with other substances to study membrane fouling (Meng et al. 2018; Dimitrios et al. 2019). Studying the membrane fouling mechanism of polysaccharides and its influencing factors can be used to better understand the fouling mechanism of membrane separation systems including MBR, allowing us to take more appropriate measures to alleviate membrane fouling.

Under the premise of increasing energy conservation, as well as emission reduction and wastewater treatment requirements, the addition of powdered activated carbon (PAC) under limited aeration conditions is considered a successful measure to mitigate membrane fouling (Ng et al. 2010; Iorhemen et al. 2017; Asif et al. 2020a, 2020b; Wang et al. 2020). As a scouring particle, PAC forms gas–solid–liquid three-phase flow under the action of aeration, proving a higher scouring effect on the membrane surface to alleviate cake layer fouling than aeration alone. However, mechanical wear of the organic membrane surface caused by PAC and a further reduction in effluent quality have been observed in membrane separation systems (Doll & Frimmel 2005; Meier 2010). Therefore, membrane separation systems composed of mechanical wear-resistant ceramic membranes have emerged, which have garnered increasing attention (Rizzo et al. 2019). Therefore, the fouling mechanism of mechanical wear-resistant ceramic membranes in microfiltration systems with simultaneous changes in TMP and J needs to be investigated.

The study attempted to clarify the factors that affect the variations in permeate flux and TMP during SA microfiltration by ceramic membranes, as well as reveal the essential relationship between filtration resistance and the filtration process in principle. SA was filtrated through a flat-sheet ceramic microfiltration membrane controlled by a peristaltic pump under three filtration conditions. (1) The influence of SA concentration on the permeate flux, TMP, and filtration resistance during SA microfiltration was studied under the condition of non-aerated-PAC with the same peristaltic pump rotation speed. (2) The influence of peristaltic pump rotation speed on the permeate flux, TMP, and filtration resistance during SA microfiltration was investigated under non-aerated-PAC and aerated-PAC conditions. (3) The influence of SA concentration and peristaltic pump rotation speed were explored on the intrinsic relationship between TMP and permeate flux during microfiltration. (4) Mathematic filtration models were used to explore the fouling mechanism.

The microfiltration SA solution under the non-aerated-PAC condition or 800 mg/L SA − 200 mg/L PAC mixture under the aerated-PAC condition was fed into the filtration tank one at a time, and the permeate was filtrated through the flat-sheet ceramic membrane using a Kamoer peristaltic pump with six rollers. The pressure generated by the filtrate through the membrane under the driving force provided by the rotation of the peristaltic pump was measured using a pressure transmitter, and recorded using a paperless recorder at a time interval of 30 s. The mass of the filtrate downstream from the peristaltic pump was measured using an electronic scale and we recorded single data points by a computer every 30 s.

The detailed analysis procedures of R_t, R_c, R_p, and R_m were described by Liu & Sun (2010). The pure water was poured into the filtration tank and we tested J₀ and TMP₀ under the driving force provided by the rotation of the peristaltic pump at speeds of 15, 30, 60, and 90 rpm. The driving force increased with peristaltic pump speed and was proportional to J₀ and TMP₀, as shown in Figure S1, in the Supplementary Materials. R_m was determined by TMP₀ and J₀ according to Equation (4). The SA solution or SA-PAC mixture was carried out under 120 min of time-to-filtration (TTF) and R_t was obtained from TMP and J after 120 min of filtration according to Equation (4). After the microfiltration experiment, SA or SA-PAC deposition on the membrane surface was carefully wiped off with a sponge, and the membrane surface was thoroughly rinsed with pure water. The pure water was then filtrated again and J_f and TMP_f were tested under the same filtration conditions. R_f, the sum of R_m and R_p, was obtained from TMP_f and J_f according to Equation (4), R_p was obtained by subtracting R_m from R_f, and finally, R_c was obtained by Equation (5).

As shown in Figure 3(c), the TMP’ in the early microfiltration stage of 200, 400, 800, and 1,200 mg/L of SA solution increased linearly with a decrease in J’. The change trend of TMP’ with J’ was expressed by the linear slope, k_SA, where the larger the absolute value of k_SA, the faster TMP’ decreased with an increase in J’. The absolute value of k_SA increased from 6.01 to 8.58 with increasing SA concentration from 200 to 1,200 mg/L (Figure 3(d)). The TMP’ in the stable microfiltration stage showed almost no changes with a decrease of J’.

The above phenomenon indicated that the early SA microfiltration stage, which was controlled by a peristaltic pump rotating at 30 rpm, was non-constant-flux and non-constant-pressure filtration, while the later stable microfiltration stage approached constant-pressure filtration. The filtrate that passed through the membrane pores was controlled by the driving force generated by the peristaltic pump, which regularly squeezed the liquid pipe. The same driving force was applied to the new ceramic membrane with the same membrane pore size and produced the same TMP₀ and J₀ (Figure S1). With fouling formation during SA filtration, a decrease in J’ and increase in TMP’ occurred simultaneously, and the changes were affected by the SA concentration. Further analysis indicated that a higher SA solution concentration was likely to produce more serious membrane fouling than the low concentration SA solution during microfiltration, resulting in faster dJ'/dt in the initial stage and lower J’ at the end of microfiltration. The microfiltration time to reach a stable stage decreased with increasing SA concentration. This phenomenon was consistent with previous observations during the initial stage of dead-end microfiltration of SA with concentrations of 20–500 mg/L, where the rotation speed of the peristaltic pump was 50 rpm (Liu 2022). A similar phenomenon was reported by Akamatsu et al. (2020), where the fluxes in the stable stage decreased due to adsorption with increasing bovine serum albumin and SA concentration in a closed-loop crossflow filtration with the same applied pressure.

In addition, J₀ was proportional to the peristaltic pump rotation speed (Figure S1), where dJ'/dt sharply increased and J’ decreased at the end of the filtration with an increase in J₀, which implied that the increase in initial permeate flux resulted in more serious fouling. A similar phenomenon was also reported by Peles et al. (2022), where the decay rate and decay amplitude of J’ at the same filtration time increased with increasing peristaltic pump rotation speed. Moreover, it should be emphasized that both J and TMP at 120 min of microfiltration increased with increasing peristaltic pump rotation speed (Figure S3b), and TMP₁₂₀ increased with J₁₂₀. Similar results were reported by Akamatsu et al. (2020), which indicated that the flux in the steady state in the closed-loop crossflow microfiltration of SA increased by 2.5 times when the applied pressure was 2.5 times that of the initial applied pressure. He et al. (2017) reported that the average TMP increased with increasing applied flux in the constant-flux crossflow filtration tests of the oil-in-water emulsions. These phenomena indicated that due to membrane fouling, TMP would increase with increasing permeate flux at any time during the filtration process, such as initial filtration time and the end of filtration time.

This study focused on the relationship between TMP and permeate flux across the entire filtration process. As shown in Figure 4(c), TMP’ increased with a decrease in J’ during the early microfiltration stage of 800 mg/L SA under the non-aerated-PAC condition. This indicated that microfiltration was non-constant-flux and non-constant-pressure filtration. Further analysis revealed that TMP’ was negatively linearly proportional to J’, and the slope of the straight line was expressed by k_SA-N-PAC. The absolute value of k_SA-N-PAC was smaller with an increase in peristaltic pump rotation speed, as shown in Figure 4(d). When the peristaltic pump rotated at lower speeds of 15 and 30 rpm, TMP’ almost did not change with J’ after 87 and 60 min of microfiltration, which implied that the later stage of microfiltration could be considered constant-pressure filtration. After SA microfiltration for 30 and 29 min controlled by the peristaltic pump rotated at higher speeds of 60 and 90 rpm, respectively, TMP’ increased sharply with a decrease in J’, implying that the later microfiltration stage approached constant-flux filtration. The different variations in TMP’ with J’ implied that the membrane fouling mechanism in the microfiltration process controlled by the high and low rotational speeds of the peristaltic pump was different, and the mechanisms for this difference were further analyzed in Section 3.4.2.

When 800 mg/L of SA was filtrated under the condition of aerated-PAC, the dJ'/dt values were in the range of 40.4–63.0% and the J’ values were 2.0–2.8 times those of SA microfiltration under the non-aerated-PAC condition at the same rotation speed. These results indicated that a decrease in permeate flux mainly occurred in the early SA filtration stage due to the formation and the development of cake layer fouling, and PAC combined with aeration could alleviate membrane fouling, resulting in a reduction of dJ'/dt and increase in J’. Similar to the changes in permeate flux and TMP with the peristaltic pump speed in SA microfiltration under the non-aerated-PAC condition, the higher permeate flux the higher TMP, which was also obtained at 120 min of microfiltration under the aerated-PAC condition, and which increased with an increase in peristaltic pump speed (Figure S4b).

The above phenomenon indicated that 800 mg/L SA microfiltration with an aeration intensity of 2.2 L/min and PAC concentration of 200 mg/L, controlled by the peristaltic pump rotating at different speeds, was non-constant-flux and non-constant-pressure filtration. Additional analysis revealed a linear increase in TMP’ with a decrease in J’ during the early SA filtration stage, while the later process (after 116, 113, 82, and 58 min of microfiltration) could be considered constant-pressure filtration, as shown in Figure 5(c). The absolute value of k_SA-PAC decreased from 6.46 to 2.67 when the speed of the peristaltic pump increased from 15 to 90 rpm, as shown in Figure 5(d). The absolute value of k_SA-PAC was 71.4–54.7% of k_SA-N-PAC during SA microfiltration under the non-aerated-PAC condition, which had the same speed as the peristaltic pump. This observation was consistent with the results presented in Section 3.1, where the lower the dJ'/dt value under the same filtration driving force, the smaller the absolute value of k. These results suggested that the PAC under aeration could significantly reduce the fouling propensity of SA (Liu 2022).

Permeate flux declination and TMP increased during microfiltration due to the increased resistance caused by pore-blocking fouling as a result of solute adsorption on the inner walls of the membrane pores and cake fouling due to foulant adsorption and deposition on the membrane surface. The resistance-in-series model easily described the relationships of different resistances with the permeate flux and TMP (Yeh & Cheng 1993). The resistance-in-series model was used in this study to investigate the possible membrane fouling mechanism of SA microfiltration under non-aerated-PAC and aerated-PAC conditions.

For 800 mg/L of SA microfiltration under the non-aerated-PAC condition, when the pump speed increased from 15 to 90 rpm, R_t increased by 1.99 times, and R_c/R_t and R_c/R_p increased from 91.99 to 98.32% and from 67.21 to 146.69, respectively, as shown in Figure 6(c) and 6(d). These results confirmed that R_c was the main contribution to R_t. With the acceleration of the peristaltic pump speed, the increases in J₀ and TMP₀ resulted in a more compressed cake layer with higher R_c and R_c/R_t.

As shown in Figure 6(e), when 800 mg/L of SA solution was filtrated under the aerated-PAC condition, R_t increased from 7.07 × 10⁹ to 1.08 × 10¹⁰ m⁻¹ with an increase in peristaltic pump speed from 15 to 90 rpm, with values of 24.5 to 18.9% for R_t during SA microfiltration under the non-aerated-PAC condition. These confirmed that the action of aerated-PAC could offer a more effective approach for membrane fouling mitigation (Zhang et al. 2019; Liu 2022). Figure 6(f) further shows that the R_c/R_p values were 3.18, 14.83, 21.53, and 26.36 times when SA microfiltration was controlled by the peristaltic pump between 15 and 90 rpm. These results confirmed that R_c was the major contributor to R_t under the aerated-PAC condition and R_c contribution increased with the acceleration of the peristaltic pump speed.

Mathematical simulation models have typically been used to understand and characterize the membrane fouling mechanism (Pradhan et al. 2014). However, models for analyzing membrane fouling for the non-constant-flux or non-constant-pressure filtration processes of ceramic membranes developed in recent years are scarce. Analysis of membrane fouling in Section 3.3 indicated that the cake layer that formed on the membrane surface caused the major component of R_t, which had to be systematically considered when evaluating the fouling mechanism.

The correlation coefficient (R²) of each model simulation curve was used to indicate the primary membrane fouling mechanism. The filtration mechanism during SA microfiltration under different conditions will be discussed in the next sections.

Figure 7(b) and 7(c) shows that the α at 120 min of microfiltration increased 1.70 times and the cake load of SA (M_SA) increased 2.01 times with an increase in SA concentration from 200 to 1,200 mg/L, respectively. Figure 7(d) further shows a high positive correlation between R_c and M_SA, indicating that increased foulant adhesion to the membrane surfaces led to increased membrane fouling (Meng et al. 2019).

From further analysis, we found that the α values were 4.50 × 10¹⁵ and 6.30 × 10¹⁵ m/kg for SA microfiltration controlled by the peristaltic pump rotating at low speeds of 15 and 30 rpm, respectively. The value of α could not be calculated during SA microfiltration controlled by the peristaltic pump rotating at higher speeds of 60 and 90 rpm. These results confirmed that SA microfiltration controlled by the peristaltic pump with a higher speed did not conform to the cake filtration mechanism. This phenomenon was likely to have been due to the blocking of pore entrances by the compressed cake layer during filtration controlled by the higher rotation speeds of the peristaltic pump. Mahamadou Harouna et al. (2019) also noted that t/V was independent of V when the cake layer was compressible. The increase in J₀ and TMP₀ resulted in a more compressed cake layer, which caused pore occlusion and corresponded to the attenuation of initial permeate flux due to higher R_c during membrane filtration. The compressed cake layer also decreased the porosity of the cake layer, which reduced R_p caused by the smaller-sized SA entering the membrane pores through the cake voids and R_p contribution to R_t.

As shown in Figure 9(b), the α values for 120 min of microfiltration decreased from 5.63 × 10¹⁴ to 4.88 × 10¹⁴ m/kg when the peristaltic pump speed increased from 15 to 30 rpm, and did not change when the speed increased from 30 to 90 rpm. The cake layer with a lower value of α had a higher permeability (Baker et al. 1985; Liew et al. 1995). It was evident that the permeability of the cake layer formed by PAC and SA increased with increasing particle size, compared to the cake layer formed by SA only. The PAC combined with aeration could be used as an effective approach for membrane fouling mitigation. The mass of the cake layer due to SA and PAC deposition (M_SA-PAC) increased from 6.52 to 19.99 mg/m², which tended to increase with the peristaltic pump speed, as shown in Figure 9(c). Further analysis indicated that for the same particle size and concentration of SA and PAC, the cake load M_SA-PAC was closely correlated with the cake resistance, as shown in Figure 9(d).

In summary, SA was a natural biological polysaccharide containing carboxyl (COO-) and hydroxyl groups (-OH), along with highly adhesive characteristics (Jin et al. 2009). More than 90% of the SA particle size was greater than 0.1 μm (diameter of the membrane pore), making the particles more likely to be retained by the membrane pores during microfiltration, forming a cake layer on the membrane surface. Increasing the SA concentration could promote the formation of larger aggregates than at lower SA concentrations (Akamatsu et al. 2020), and they could adhere to the membrane surface and block the membrane pores due to adsorption, resulting in a higher fouling rate and lower permeate flux. At the same concentration, SA formed a more compact cake layer with higher cake resistance during microfiltration controlled by the peristaltic pump rotating at higher speeds, compared to lower speeds. PAC played two roles in mitigating membrane fouling under the aeration condition. First, as scouring particles fluidized under the action of aeration, PAC induced surface shear and detached the accumulated membrane fouling substances (Wang et al. 2020). Second, the cake layer formed by SA and PAC had a much looser structure with lower cake resistance compared to the one formed by SA alone during microfiltration. The structure of the PAC-mediated SA cake layer was compressed with the peristaltic pump speed and the cake void was smaller, resulting in an increase in the amount of SA and PAC intercepted, as well as a higher cake resistance contribution to the total resistance.

In this study, we investigated the effects of SA concentration and peristaltic pump rotation speed on membrane fouling during SA microfiltration and drew the following conclusions.

SA microfiltration controlled by the peristaltic pump was non-constant-flux and non-constant-pressure filtration, which was independent of the operating conditions. The membrane fouling that formed in the initial filtration stage blocked the membrane pore entrances, resulting in a rapid decrease in J’, at which point TMP’ increased. The variations in TMP’ with J’ were affected by the SA concentration and peristaltic pump speed, which further affected the change in TMP’ with J’ and the resistance in the later filtration stage.

In the microfiltration of the SA solution under the non-aerated-PAC condition with a peristaltic pump rotation speed of 30 rpm, the fouling mechanism in the early filtration stage followed the cake filtration model, where the level of increase of TMP’ with the decrease of J’ increased with higher cake resistance due to a higher amount of SA accumulation on the membrane surface when the SA concentration increased. The later filtration stage approached constant-pressure filtration, and a lower J’ and higher TMP’ were obtained due to more serious fouling with an increase in SA concentration.

In the early stage of 800 mg/L of SA microfiltration controlled by the peristaltic pump, membrane fouling was well described by the cake filtration model at lower rotation speeds of 15 and 30 rpm, while the complete blocking model was suitable for higher rotation speeds of 60 and 90 rpm. The trend of TMP’ increasing with the decrease of J’ slowed down with an increase in peristaltic pump speed. In the later stage of SA microfiltration, filtration of the peristaltic pump at low speeds of 15 and 30 rpm approached constant-pressure filtration, while filtration at high speeds of 60 and 90 rpm approached constant-flux filtration.

In the early stage of 800 mg/L of SA microfiltration under the aerated-PAC condition, the fouling mechanism was well described by the cake filtration model. With an increase in peristaltic pump speed, the resistance of the cake layer formed by SA and PAC increased due to an increase in the amount of SA and PAC attached to the membrane surface. The trend of TMP’ increasing with a decrease of J’ in the early stage of microfiltration slowed with an increase in peristaltic pump speed. The later filtration stage approached constant-pressure filtration. Due to the scouring of PAC, the total resistance was reduced by 24.5 to 18.9%, compared to without aerated-PAC, respectively, when the rotation speed of the peristaltic pump increased from 15 to 90 rpm.

Studying the factors that influenced the transmembrane pressure, permeate flux, and fouling mechanisms was helpful for optimizing the operating conditions of the microfiltration process. In future work, we will investigate the factors that affect the transmembrane pressure and permeate flux, as well as the reversible and irreversible resistance in the microfiltration of other polysaccharides, or mixtures of polysaccharides and other organic or inorganic substances in the microfiltration process.

The author acknowledges financial support from the Shanxi Scholarship Council of China (2017-108).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.