A new concept of a rotating hollow fibre membrane module:impact of rotation on fine-bubble aeration

The use of membrane bioreactor (MBR) systems becomes a growing interest due to tighter environmental regulations, water scarcity and the rising need of water reuse (Judd 2016; Paul & Jones 2016). The MBR system offers the ability to achieve high nutrient removal efficiency and complete biomass retention at a small footprint (Hoinkis et al. 2012; Meng et al. 2017). However, MBR systems also suffer from membrane fouling and clogging, which impacts the performance of the system significantly and makes frequent cleaning procedures inevitable. In addition, MBR systems are still considered expensive in comparison to conventional activated sludge (CAS) systems. The average energy consumption of immersed MBRs per m³ of treated municipal wastewater ranges between 0.8–1.1 kWh/m³ with the lowest reported value being 0.4 kWh/m³ (Judd 2011; Krzeminski et al. 2017).

For both MBR and CAS systems, aeration is responsible for more than 50% of the energy demand (Judd 2011; Helmi & Gallucci 2020). The overall aeration demand is typically higher for MBR than for CAS systems because the higher sludge concentration (floc/solid holdup) lowers the oxygen transfer rate and an additional air cross flow is required to reduce membrane clogging and fouling (Henkel et al. 2011).

Several studies were conducted to improve the performance of MBRs in terms of fouling reduction and aeration efficiency (Judd 2016; Helmi & Gallucci 2020). One possible method is the introduction of a dynamic shear force through rotation (Jaffrin 2008; Wu et al. 2008; Jiang et al. 2013; Ruigómez et al. 2016; Zsirai et al. 2016). Rector et al. (2006) did experiments on a rotating membrane bioreactor using a hollow fibre (HF) module. They concluded that the rotation of HF membrane modules provides an efficient method of prompting turbulent flow and higher dispersion in an MBR system. Paul & Jones (2016) also showed that rotating MBR systems had better performance than static MBR when both systems were operated under similar conditions.

This paper introduces a novel membrane configuration where horizontally aligned HFs are connected to a vertical permeate tube. The membrane module rotates in the activated sludge reactor. The rotation creates a permanent shear force at the fibre and eliminates the need of a uniformly distributed aeration device below the membrane unit. In comparison to current membrane systems, the fibre length is relatively short (<20 cm), which is expected to improve the backwash and cleaning process significantly.

The study was conducted to investigate the impact of different rotation velocities on oxygen transfer in a CAS plant. Hence a fine bubble aeration system and activated sludge from a nearby conventional activated sludge plant were used. To determine the impact of sludge flocs on aeration performance also clean water tests were performed to evaluate the ratio of sludge to clean water oxygen transfer coefficients (α-factor). Finally, the rotation energy was calculated and put into relationship with the oxygen transfer measurements.

The membrane module (Figure 1) consists of 1,950 HF membranes with one side potted horizontally into the permeate tube and sealed at the other side. The material of HF membrane is polyvinylidene difluoride (PVDF) with the pore size of 0.3 μm. The total surface area of the membrane module is 2.25 m², which leads to a packing density of 81 m²/m³. Due to the horizontal membrane alignment and the applied rotation, all membranes are in contact with the air bubbles even though only one segment of the membrane prototype is aerated.

The measurements were carried out in a transparent tank filled with clean water (softened tap water) and activated sludge with a volume of 1 m³. The softened water was chosen to prevent any hardness precipitation during the experiments. The activated sludge was obtained from the wastewater treatment plant in Königsbach, Germany (55,000 PE). The experiments were performed in batch mode. No feed or filtrate was added nor taken during the experiment. A Flexnorm 500 diffuser (OTT System GmbH & Co. KG, Langenhagen, Germany) was used as fine bubble air diffuser. The airflow rates and dissolved oxygen (DO) concentrations were measured with a thermal flow sensor TA Di 21.6 GE (Hoentzsch GmbH, Waiblingen-Hegnach, Germany) and three oxygen electrodes (PRONOVA Analysentechnik GmbH & Co., Bad Klosterlausnitz Germany), respectively. In addition, the ambient parameters (air temperature, pressure and partial humidity) were documented through a weather station. Figure 2 shows the setup of the pilot plant.

Tables 1 and 2 below show the sludge characteristics and an example of the floc sedimentation test to define the hydrostatic floc volume (HFV) for the activated sludge during the experiment days. After an initial decrease in mixed liquor volatile suspended solids concentration (MLVSS) and HFV during the first 12 hours following activated sludge collection, both values remained relatively stable during experiments, indicating that their impact on the measured oxygen transfer coefficient was stable.

As the oxygen concentration in the sludge ranged from 7.2 to 9.9 mg/L and the saturated concentration was up to 10.13 mg/L O₂ during the experiments, the off-gas method was not appropriate in these conditions (Krause et al. 2003). Therefore, the desorption method described by Wagner et al. (1998) and DWA-M 209 (2007) was chosen for calculating the oxygen transfer coefficient (k_La) value. In this method, the oxygen transfer rate is determined from the decrease in the DO concentration when the water is diffused with normal air. The DO concentration in clean water and in activated sludge was artificially increased above saturated concentration in advance using pure oxygen aeration into the water. Five different rotational speeds (0, 10, 30, 40 and 50 rpm) for clean water and three different rotational speeds (0, 30 and 50 rpm) for activated sludge were tested at three different airflow rates (1, 2, 4 m³/h). The k_La value was normalized to the standard conditions (k_La₂₀) at a water temperature of 20 ̊C and an atmospheric pressure of 1,013 hPa, and with the correction factor for a salt concentration of 1 g/L, due to significant differences between the salt content of clean water and wastewater (DWA-M 209 2007).

During activated sludge experiments, it is compulsory to maintain constant respiration rates. This was achieved by aerating the activated sludge for at least 12 hours before the experiment started. This nullified the impact of degradable surfactants on oxygen transfer. Consequently, the results should mainly reflect the impact of activated sludge flocs on oxygen transfer depletion.

Three oxygen sensors recorded the change of the oxygen concentration in the sludge at a constant airflow rate and a constant sludge concentration. From these records, the k_La was determined by non-linear regression and an average k_La was calculated. Subsequently, the airflow rate was changed and the procedure was repeated at the same sludge concentration. Three airflow rates were chosen for each experiment series with the same sludge concentration. Afterwards, the three average k_La₂₀ values were plotted against the specific airflow rate and trends were defined by polymeric trendlines.

This procedure was performed in clean water and activated sludge. Finally, the comparison between the k_La₂₀ value of activated sludge and clean water defined as the α-factor was calculated by dividing the trendline equation at a certain sludge concentration by the equation obtained during the clean water experiment. As x-variable, the three applied specific airflow rates were inserted into the equation and as a result, three α-factors were received for each sludge concentration. This modus operandi was applied since it was not possible to have identical airflow rates for each experiment series. Humidity, air pressure, air temperature and the air blower influenced the airflow rate, which was transformed to standard conditions.

The energy consumption was calculated from the electrical demand of the blower and the reading from the rotational device. The blower energy consumption followed the equation introduced by the blower manufacturer: Power (W)=35×volume flow. According to the reading from the rotational device, it consumed 16.52 W at 30 rpm and 17.27 W at 50 rpm.

The results in clean water showed that both specific airflow rates and the rotational speeds of the membrane prototype influenced the k_La₂₀ value in direct proportion, see Figure 3.

The oxygen transfer rate rises steadily because with increasing airflow rate, the gas holdup increases. As the airflow rates increase, the efficiency decreases slightly due to the creation of bigger bubbles at the flexible orifice and bubble coalescence nearby the diffuser, which is typical of membrane diffusers (Painmanakul et al. 2004). Most likely, the effect of coalescence is additionally favoured through the contact of the bubbles with the rotating membrane module, which acts like a barrier for the uprising bubbles and facilitates the collision of the bubbles. As the bubble size was not investigated in this study, there is no final proof of this assumption.

Furthermore, the rotation of the membrane module leads to an improvement in oxygen transfer, as shown in Table 3. With the exception of the 10 rpm experiment, every increase of the rotational speed by 10 rpm raised the oxygen transfer by roughly 10%. The biggest improvement was achieved at the highest specific airflow rate of 4 m³/(m³·h) and 50 rpm with an increase in k_La₂₀ of 52% compared to the experiment without rotation. With increasing airflow rate, the improvement in oxygen transfer rate decreased slightly for the same rotational speed except for the experiment at 50 rpm where it increased from 43% to 52%.

The following observations of the effect of rotation of the membrane module were made:

• At a rotational speed of 10 rpm, no effect on bubble formation at the orifice was visible. The bubbles rose straight to the top of the reactor. With increasing rotational speed, this pattern changed. Figure 4 shows that at a rotational speed of 30 rpm, the increased fluid force led to an angular raise of the bubbles, which was even more pronounced at 50 rpm. The increased fluid flow force at the orifice stimulated a faster bubble detachment and clearly changed the bubble raise behaviour. Both effects had a positive impact on oxygen transfer because the interfacial area and the residence time of the bubbles were increased.

• A second effect was visible at the level of the membrane device. At 0 and 10 rpm, the bubble entered the membrane device and little to no effect on the bubble rising behaviour was visible. With increasing rotational speed, the bubbles experienced an additional horizontal acceleration which led to an increase in the bubble residence time and improved oxygen transfer. However, at a high rotational speed (50 rpm), the effect of horizontal acceleration during bubble formation became the determining factor, so the majority of bubbles did not rise through the membrane device but bypassed the module. On this account, the enhancement of k_La₂₀ with increasing airflow rate at 50 rpm was caused by the reduction of coalescence phenomena triggered by the rotating membrane module. This effect should not prevent the primary intention of crossflow aeration to provide sufficient shear forces to membrane fibres to reduce fouling/clogging of the module. So far the question is open whether the higher rotational speed can compensate/offset the need for an additional air crossflow to suppress fouling and clogging of the module.

Additional experiments were performed to investigate the effect of the rotating membrane module on oxygen transfer in activated sludge from a municipal wastewater treatment plant.

Similarly to the results obtained in clean water, the results of the oxygen transfer experiments for activated sludge at 0, 30 and 50 rpm also showed that k_La₂₀ values improved in direct proportion as both specific airflow rates and the rotational speeds increased (Figure 5). Similar improvement rates caused by rotation as in clean water were measured as well (Table 4) due to the same dependencies provided in the activated sludge system.

Table 5 shows the obtained α-factors at different specific airflow rates. The α-factors ranged from 0.77 to 0.81, almost independently from rotational speeds and specific airflow rates.

A slight increase with rotational speed can be observed for specific airflow rates of 1 and 2 m³/(m³·h), while with 4 m³/(m³·h), no connection to rotational speed was detected. This observation suggests that the effect that was responsible for the decrease in oxygen transfer by around 20% in activated sludge must act nearly independently from the rotational speed and the applied airflow rate.

Germain et al. (2005) and Krause (2005) argued that one effect that impacts oxygen transfer depletion in activated sludge could be attributed to the increase in apparent viscosity. With this assumption, an improvement in the α-factor with increasing rotational speed can be expected. Because activated sludge is described as a non-Newtonian pseudoplastic fluid (Rosenberger 2003; Yang et al. 2009), an increasing shear stress should have decreased the apparent viscosity of the sludge and thus improved the α-factor significantly. Such an effect could not be observed during experiment series with the rotatable membrane module system.

Several other authors (Wagner & Pöpel 1996; Rosso et al. 2006; Garrido-Baserba et al. 2020) argued that dissolved organic matter such as surfactants mainly contribute to the depletion of the α-factor in activated sludge. As in the case of apparent viscosity, a positive impact of rotation on the alpha factors can be expected if surfactants are the major contributors to the decrease observed. The higher turbulence introduced by the rotation, which clearly led to an improvement in oxygen transfer should have also increased the surface renewal rate of the bubbles and as such decreased the effect of surfactants significantly. Again, such an effect could not be observed during experiment series with the rotatable membrane module system, particularly because the activated sludge was stabilized before the oxygen transfer tests.

Another factor that influences oxygen transfer in a three-phase system is the floc/solid holdup (Deckwer 1992; Mena et al. 2005; Henkel et al. 2011). Already van der Kroon (1968) demonstrated that aluminium hydroxide and activated sludge flocs steadily decreased oxygen transfer with increasing concentration. Henkel (2010) demonstrated that despite different reactor configurations (bubble column, airlift reactor), diffuser systems (fine bubble, coarse bubble, combination of both) and sludges/slurries of different origins (e.g. iron hydroxide flocs, municipal activated sludge, greywater activated sludge), the α-factor showed similar values if the floc volume was used as comparative parameter. Because the different rotational velocities did not show a significant impact on the obtained α-factors, the floc volume seems to be the main driver of oxygen transfer depletion according to present study. The rotational speed indeed impacted the bubble formation and gas holdup, which lead to an increase in oxygen transfer compared to the results obtained without rotation, but it did not impact the reduction of the interfacial area of the bubble caused by the sludge flocs.

To investigate if the energy demand required by the rotation of the membranes could be compensated by the improved oxygen transfer rate, the standard aeration efficiency (SAE) was plotted against the specific airflow rates as shown in Figure 6.

Except for clean water experiments at the lowest airflow rate, the SAE was equal to or higher than the experiments without membrane module rotation. An improvement of 24% at 2 m³/(m³·h) and 36% at 4 m³/(m³·h) was achieved at a rotational speed of 50 rpm if matched against the activated sludge results at 0 rpm. Consequently, the additional energy required for rotation was overcompensated by the improved oxygen transfer efficiency in the activated sludge experiments.

A set of experiments was conducted with a new rotating type of membrane module suitable for membrane bioreactor applications to study its effect on oxygen transfer and energy consumption. The rotation of the membrane module showed the following effects:

• oxygen transfer was significantly improved by the application of rotation (∼50% at 50 rpm),

• the rates of improvement in oxygen transfer for clean water and activated sludge were similar,

• the α-factors showed comparable values for all experiments independent of the rotational speed,

• an energy comparison indicates that the additional energy demand for rotation can be compensated by the better oxygen transfer efficiency,

• at the highest rotational speed (50 rpm), the rotation induced such high shear forces to the bubble formation that the majority of bubbles bypassed the membrane module,

• compared to full-scale systems using fine bubble aeration, the SAE was low (<1,2 kgO₂/kWh vs. 4 kgO₂/kWh), which can be attributed to the pilot-scale blower and motor for rotation. Consequently, a transfer of these results to full-scale applications is not feasible and requires further investigations.

The proposed membrane module design encourages the possible reduction of total membrane area needed in comparison to existing membrane design available in the market. According to Lo et al. (2015), who studied cost estimation for small membrane bioreactor, the membrane cost contributes to approximately 50–64% to total capital expenditure (CAPEX). Therefore, if the proposed membrane module design can reduce the total membrane area needed by 50%, it has potential to reduce total CAPEX by approximately 30%.

Furthermore, the challenge for a wider use of MBR technology is its integration into an existing CAS system without major modification. The membrane module design in present study shows high feasibility to meet these expectations. Further experiments include the testing of the rotating module under process conditions (filtration, cleaning, fouling behaviour) and the study of the impact of rotation to a coarse bubble aeration system at high sludge concentrations.

Fathul Mahdariza: Conceptualization, Methodology, Investigation, Data Curation, Writing, Review and Editing; Ignacio Domingo Rimoldi: Conceptualization, Methodology, Investigation; Jochen Henkel: Conceptualization, Methodology, Validation, Writing, Review and Editing; Tobias Morck: Conceptualization, Methodology, Validation, Supervision, Writing, Review and Editing

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The authors want to thank the Ministry of Finance of the Republic of Indonesia for the LPDP scholarship provided to Fathul Mahdariza (Ref. Number: S-1289/LPDP.4/2018). The research did not receive any other specific grants from funding agencies in the public, commercial or not-for-profit sectors. Moreover, the authors want to thank deeply the following people at Institute for Water and River Basin Management (KIT, Germany) for valuable technical and laboratory assistance: Sven Urbach, Tobias Mumbauer, Manfred Lösche, Patrick Dufner and Maik Erhard.