Evaluation of microporous hollow fibre membranes for mass transfer of H2 into anaerobic digesters for biomethanization

BACKGROUND: With high surface-to-volume ratios, hollow fibre membranes offer a potential solution to improving gas–liquid mass transfer. This work experimentally determined the mass transfer characteristics of commercially available microporous hollow fibre membranes and compared these with the mass transfer from bubble column reactors. Both mass transfer systems are considered for biological methanization, a process that faces a challenge to enhance the H2 gas–liquid mass transfer for methanogenic Archaea to combine H2 and CO2 into CH4. RESULTS: Polypropylene membranes showed the highest mass transfer rate of membranes tested, with a mass transfer coefficient for H2 measured as kL = 1.2×10−4 ms−1. These results support the two-film gas–liquid mass transfer theory, with higher mass transfer rates measured with an increase in liquid flow velocity across the membrane. Despite the higher mass transfer rate from polypropylene membranes and with a liquid flow across the membrane, a volumetric surface area ofα = 10.34 m−1 would be required in a full-scale in situ biological methanization process with much larger values potentially required for high-rate ex situ systems. CONCLUSIONS: The large surface area of hollow fibre membranes required for H2 mass transfer and issues of fouling and replacement costs of membranes are challenges for hollow fibre membranes in large-scale biological methanization reactors. Provided that the initial bubble size is small enough (de <0.5 mm), calculations indicate that microbubbles could offer a simpler means of transferring the required H2 into the liquid phase at a head typical of that found in commercial-scale anaerobic digesters. © 2019 The Authors. Journal of Chemical Technology & Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.


Introduction
Gas-liquid mass transfer rates are important in a wide variety of chemical and biological processes and mass transfer design needs to consider a range of factors specific to the application, including the scale and dimensions of the reactor, sensitivity of biological cells to shear forces, reaction rate and process economics. An important development in meeting these requirements is the application of new membrane materials configured as novel types of gas diffuser.
These have the potential to increase gas-liquid mass transfer while avoiding the use of energy-intensive mixing systems and high flow rates in the gaseous phase, both of which can affect performance and increase operating costs.
One such process that depends on gas transfer is the biological methanisation of CO 2 . This reaction has attracted considerable commercial interest recently, as it offers a possible route to energy storage via the conversion of renewable electricity into H 2 and then to CH 4 [1]. The conversion process utilises hydrogenotrophic methanogenic Archaea [2], and may be conducted in-situ within an anaerobic digester, with H 2 added to combine with CO 2 in the biogas [3,4]; or ex-situ in a separate hydrogentrophic reactor, using gaseous feedstocks of H 2 and CO 2 [5,6] or H 2 and biogas [7]. Ex-situ reactors typically have higher volumetric conversion rates from H 2 /CO 2 to CH 4 than in-situ reactors [8]. Biological gas upgrading by either of these methods has the potential to reduce both the costs and the methane slippage characteristic of existing physico-chemical biogas upgrading technologies [9], and may also provide a future tool for carbon capture and utilisation.
Due to the low solubility of H 2 and the reaction stoichiometry (4:1; see equation 1), gas-liquid mass transfer is the limiting step for this process and as such is recognised as a major engineering challenge [1]. Research work conducted at laboratory and pilot scale has experimented with a range of different mass transfer approaches for H2 into biological methanisation reactors. Given the mainly pre-commercial stage of this technology the most effective mass transfer process for full-scale systems has yet to be determined, and mass transfer approaches used in laboratory scale experiments may not be the most suited to full-scale operation.
The range of reactor types that has been tested for biological methanisation includes continuous stirred tank reactors (CSTR) [10], fixed bed reactors [6] and hollow fibre membrane bioreactors [5]. Kougias  To improve mass transfer from bubbles, previous researchers have also used Accepted Article impellers to increase liquid phase mixing and turbulence [11,12,13,14,15]. For hollow fibre membranes to enhance gas-liquid mass transfer during biomethanisation correct selection of membrane material, characteristics and system configuration is essential. In the current work mass transfer coefficients were compared for four different commercial hollow fibre membranes, with the effect of gas species, gas flow rates, gas pressure in the membrane lumen and liquid crossflow velocities considered for a selected membrane type. Experimental work was conducted using tap water to compare the physical mass transfer effectiveness of different membrane systems, without the added complexities and dynamics of a biological system. Additionally, the reactor system as a whole needs to be considered from the viewpoint of scale-up, which will affect a number of gas transfer parameters. The performance of the most suitable microporous hollow fibre membrane tested was thus compared with simulated results for bubbled systems at an operational scale.

Accepted Article
This article is protected by copyright. All rights reserved.

Mass transfer calculations
The mass transfer rate is a critical design parameter for gas-liquid mass transfer systems, such as hollow fibre membrane contactors. The mass transfer rate is defined as the change in concentration with time and is shown in equation 2 for liquid side controlled mass transfer. the calculation methodology detailed previously by Mendoza et al. [18].  [20] or Ahmed et al. [21].
Mass transfer theory indicates that a species transferring from the gas phase across a hollow fibre membrane to the liquid phase is required to overcome resistance from the gas side and the liquid side as well as the membrane [22]. Non-wetted membrane pores: Wetted membrane pores: In this case k G , k L and k M are the gas, liquid and membrane coefficients, respectively, while H (mol L −1 atm −1 ) is Henry's law coefficient of the gas species in the liquid phase, listed in Table 1 for the gases tested in this work.
The solubility of H 2 in water is very low, thus H 2 has a large Henry's law coefficient, and this results in the resistance to mass transfer being significantly larger on the liquid side than the gas side. The magnitude of the membrane resistance k M is dependent on the degree of wettedness of the membrane pores [23].   The hollow fibre membranes were potted with epoxy resin into a linear configuration inside two acrylic fixings with gas inlets/outlets connected at each end. Figure 1c shows an image of the 3 mm hollow fibre membranes. Before each experiment gas-impermeable sampling bags were filled with the desired gas (O 2 , H 2 , and CO 2 ; BOC, UK), which was then pumped into the hollow fibre membrane fixing with a peristaltic pump (Watson Marlow, UK). When operating at atmospheric pressure the output gas was collected in an outlet gas sampling bag. For experiments with a pressurised gas phase the inlet gas line was divided in two, with both inlet lines connected to the hollow fibre fixings.

Accepted Article
A pressure transducer (Hydrotechnik, UK) was connected to the inlet gas line to measure the gas pressure. The hollow fibre fixing was located in the centre of the reactor, and could be positioned in different configurations. Tap water was used as the liquid phase, which was recirculated with a centrifugal pump (CEB103, Clarke, UK). During CO 2 experiments the pH was measured using a glass bulb pH electrode (Extech, UK) calibrated with buffer solutions at pH 4 and pH 7. DO and

Accepted Article
This article is protected by copyright. All rights reserved. for two hours to ensure the gas concentration in the headspace had equilibrated.
A 10 mL gas sample was then extracted with a syringe and analysed using a gas chromatograph as detailed above.    This corresponds with results from previous studies, such as Yang and Cussler [19], Ahmed et al. [21] and Zhang et al. [28]. Considering the two-film theory noted that using a hydrophilic membrane, in which the pores will be filled with water, leads to a higher membrane resistance, and therefore the liquid side mass transfer resistance is of less importance for the overall mass transfer rate [19].

Results & Discussion
These experiments were repeated for the mass transfer of CO 2 and H 2 across the polypropylene hollow fibre membrane (P P # 2). Table 3 shows the mass transfer coefficients recorded with a liquid flow velocity of u L = 6.5×10 −3 m s −1 across the membrane. The overall liquid side mass transfer coefficient for H 2 is slightly higher than that of O 2 , due to the higher liquid diffusivity of H 2 [29]. Table 3: Overall liquid side mass transfer coefficient for O 2 , CO 2 and H 2 through polypropylene hollow fibre membrane (P P # 2)

Gas
Overall liquid side mass transfer coefficient (K L ) The overall liquid side mass transfer coefficient taken in this example de-  Accepted Article and maintenance costs [30].

Comparison with bubbled systems
An important advantage of using hollow fibre membranes for gas-liquid mass transfer is the complete transfer of the gas species on crossing the membrane.   The experiments in this work were carried out at room temperature, between 20 − 25 o C, which is lower than either the mesophilic (35 o C) or thermophilic (55 o C) temperatures at which anaerobic digesters normally operate. There is a slight reduction in H 2 solubility at higher temperatures, which will therefore reduce the concentration driving force for mass transfer. The presence of surfactant substances is known to have an effect on mass transfer rates from bubbles [32] and these are likely to be present at much higher concentrations in digestates or in the pure cultures of ex-situ biogas upgrading reactors, than in tap water; but the scale of this effect is unknown. Using tap water in the experimental work also prevented fouling of the membrane, ensuring consistent membrane conditions throughout the experiments. In the absence of more data, tap water was therefore considered an acceptable liquid medium for testing purposes.
The mass transfer coefficients used have been calculated based on the theories of mass transfer for a 'mobile' gas-liquid interface taken from Higbie [35] and Montes et al. [36] and an 'immobile' gas-liquid interface from Frössling [37].

Accepted Article
This article is protected by copyright. All rights reserved.   Different options are available with respect to microbubble diffusers. Often an elevated pressure (3.5 bar) is required to produce the microbubbles; in recent research, however, microbubbles have been produced at a reduced input gas pressure and therefore lower energy requirement [39]. The production of  Accepted Article microbubbles may also reduce fouling on the gas-liquid membrane, which may be an issue with bubble-less mass transfer from hollow fibre membranes [40].

Comparison with different membrane systems
As shown in section 3.1, a specific membrane area of 10.3 m 2 m −3 is required to supply H 2 for the in-situ conversion of endogenously-produced CO 2 in a conventional anaerobic digester with a volumetric biogas production rate of 1 m 3 m −3 day −1 at 50% CO 2 content. The performance of current membranes therefore appears to be adequate at volumetric biogas production rates typical of conventional digesters treating organic wastes or municipal wastewater biosolids. Table 6 gives examples of the use of membranes in anaerobic membrane bioreactors (AnMBRs) for the treatment of liquid effluents, and in biomethanisation studies. Based on these, provision of the specific membrane area for H 2 mass transfer found in this study is feasible. Martin-Garcia et al. Other studies utilised considerably higher specific membrane areas, from 62 to 3208 m 2 m −3 [13,5,44,45,46]. The higher values for specific membrane surface areas reported in Table 6 reflect the laboratory scale of these studies, and these higher membrane areas maybe less suited to scaling up to pilot and full scale reactors.
Although the membrane area determined from this work is feasible for insitu CO 2 conversion, current and forthcoming developments in the field of Accepted Article biomethanisation mean that other types of system may require much higher volumetric gas transfer rates. B et al. [47] demonstrated it was feasible to convert the biogas from three conventional digesters within a single unit in a combined in-situ and ex-situ system. In high-rate ex-situ systems, volumetric methane production rates of 40 m 3 m −3 reactor day −1 have been achieved at laboratory scale [48], requiring gas inputs of at least 160 and 40 m 3 m −3 day −1 of H 2 and CO 2 , respectively. Methane evolution rates of over 500 L L −1 day −1 have been demonstrated in short-term operation in chemostat cultivation of pure cultures [2]. If biomethanisation of CO 2 is to make a significant contribution to carbon capture, volumetric methane productivities of this order will be needed in order to minimise reactor size, with correspondingly high rates of mass transfer.
A municipal incinerator with a single grate, for example, may produce around 100 000 m 3 CO 2 day −1 , while a large cement kiln can produce up to 2.5 million m 3 CO 2 day −1 . A biomethanisation reactor with a volumetric conversion rate of 500 m 3 m −3 day −1 may be a feasible option for the future, but the associated mass transfer rates may require significant advances in membrane performance.
In either case, the use of microbubbles may be preferable when the head of liquid is sufficient to allow optimal H 2 transfer to the liquid, in order to avoid operational problems caused by an increase in pressure resistance due to membrane fouling. Luo et al. [45] reported that the biofilm accumulation on the membrane surface restricted H 2 diffusion and thus consumption. Consequently, the use of membranes for gas transfer may be more suitable in reactors with low liquid head depth, such as pilot plants or laboratory-scale reactors.
This could help smaller units maintain a mass transfer performance similar to industrial plants, or as an alternative the H 2 -diluted headspace gas could be recirculated through the reactor, providing increased mixing as well as gas-liquid mass transfer.

Accepted Article
This article is protected by copyright. All rights reserved.

Conclusion
Biomethanisation utilising hydrogenotrophic methanogens to convert electrolytically produced H 2 with CO 2 from biogas to form CH 4 is a promising approach for energy storage. One current engineering challenge is to enhance the H 2 gas-liquid mass transfer for the methanogens to produce CH 4 . This work has characterised the gas-liquid mass transfer from microporous hollow fibre membranes and analysed the potential for scale-up within an anaerobic digester. The large surface area of hollow fibre membranes required may make this approach less attractive, particularly considering issues of fouling and replacement costs of the membranes. Alternatively microbubbles could provide the necessary H 2 into the liquid phase given a sufficient liquid head typical of that found in commercial-scale anaerobic digesters.