Influence and use of gas bubbles for the reduction of stray currents in industrial alkaline water electrolysis

EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 1/8 A1802 Influence and use of gas bubbles for the reduction of stray currents in industrial alkaline water electrolysis Simon Appelhaus*, Xinming Chen, Maik Becker, Thomas Turek Clausthal University of Technology, Clausthal-Zellerfeld/Germany; *Contact corresponding authors: www.EFCF.com/ContactRequest Abstract Alkaline Water Electrolysis (AWE) is a key technology for green hydrogen production. It is named after the highly conductive alkaline electrolyte, typically ~30 wt.% KOH solution, that is pumped through the electrolysis cells during operation. In industrial applications, many individual cells are electrically connected in series to form a "stack". Because the electrolyte is conductive, a short circuit occurs and some current bypasses the cells through the manifold, known as a stray, leakage or shunt current. This current has several undesirable effects: a reduction in current efficiency, a maldistribution of load across the cells and corrosion due to electrochemical reactions outside of the electrolysis cells [1]. One approach to increase the resistance and lower stray currents in the electrolyte is to introduce gas bubbles into the distribution tubes, decreasing the area of the conductive liquid. This occurs naturally at the outlet of the stack, where the produced gas and the electrolyte leave the cells mixed. It is usually assumed that this significantly increases the resistance, especially when a plug flow or annular flow regime is reached. However, no experimental studies have been carried out to quantify this resistance increase. In this work, resistance in circular tubes at different gas and liquid flow velocities is measured at industrial AWE conditions. Furthermore, the transfer of this concept to the feed distribution via the addition of gas bubbles to the electrolyte is evaluated. Figure 1: Example of liquid-gas flow regimes with increasing gas fraction: a) dispersed, b) bubbly, c) slug, d) annular [2]. [1] A. T. Kuhn, J. S. Booth, J Appl Electrochem. 1980, 10 (2), 233–237. DOI: https://doi.org/10.1007/BF00726091. [2] L. Chen, Y. S. Tian, T. G. Karayiannis, International Journal of Heat and Mass Transfer. 2006, 49 (21–22), 4220–4230. DOI: https://doi.org/10.1016/j.ijheatmasstransfer.2006.03.025. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 2/8 Introduction Alkaline water electrolysis (AWE) is currently receiving significant interest in both research and industrial development. It is one of two electrolysis technologies that have reached technical maturity, and there are multiple large-scale production facilities for AWE stacks in operation, under construction, and in planning [3]. A key problem during the scale-up of AWE systems is the occurrence of stray, shunt or leakage currents. These currents occur only at an industrial scale in stacks with many cells connected to a common electrolyte distribution system, typically a manifold connected to the cells with individual channels. Due to the higher voltage differences across the entire stack, significant current can short-circuit through the electrolyte channels outside the cells. The consequence of this stray current is a reduction of Faradaic efficiency, the production of contaminant gas in the cells and corrosion on grounded parts in contact with the electrolyte [1]. This work deals with the impact gas bubbles in the connection channels have on the stray current quantity. 1. Scientific Approach Since reducing the electrolyte conductivity or the number of cells is not desirable for industrial operation, minimizing stray currents is typically realized by increasing the channel resistance outside the cells. The most common approach is using long and thin electrolyte paths, for example, thin individually connected tubes, which also increase the system's pressure drop and require significant space and design compromises [4–6]. In this work, the influence of bubbles blocking the electrolyte path and their influence on the resistance are evaluated as an alternative way for stray current reduction. Bubbles are already present in the two-phase flow of the outlet channels of the electrolyzer, and it can be assumed that they have a significant impact on the outlet resistance. It may also be beneficial to add gas into the feed of the electrolyzer to significantly increase the resistance in these channels without simultaneously increasing the pressure drop. This approach has been detailed in two patents issued in 1970 [7, 8], but the patents have elapsed, and the method has not been tested in scientific literature. While the impact of bubbles inside the electrolyzer has been studied extensively in recent years [9–16], there is no research on their impact on stray currents available in the literature. The influence of bubbles inside the small tube is hypothesized to be different than in the bulk of the electrolysis cell, as the bubble size is much larger relative to the cross-sectional area of the tube. Some studies measure stray currents in a laboratory setting, but these typically focus on the purely liquid inlet side of the electrolyzer [17–20]. Similarly, simulations of stray currents disregard the outlet tubes or the influence of the bubbles [4, 21, 22] or use the Bruggeman two-phase model [23, 24]. The Bruggeman correlation for spherical dilute inclusions has been used successfully to predict the two-phase fluid conductivity inside the cell [25, 26]. However, it has three conditions which are violated to a different extent by the bubble-tube system: (i) negligibly small particle size compared to the observed volume, (ii) unordered particles and (iii) no inclusion of one substance in the other. Especially assumption (i) is evidently not valid if large, agglomerated gas bubbles are inside the tube. In this research, the resistance increase of the system at different gas-to-liquid ratios is evaluated to better understand the impact of bubbles on the stray currents in real alkaline water electrolysis systems. The significant variations inherent in these measurements are analyzed, and a reliable data processing method is developed. In addition, the size of the bubbles is varied at the same gas flow to investigate the influence of the flow regime on the resistance. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 3/8 2. Experiments The experimental setup was designed with a direct comparison to an industrial system in mind, and the conditions were chosen accordingly. Figure 2: Simplified process flow diagram of the experimental setup. Figure 2 depicts the entire experimental setup. 30 ± 2 wt.% KOH electrolyte is pumped from the 1 L storage tank (T01) by a peristaltic pump at a constant flow rate of 500 mL/min through a pulsation dampener (T02) and a heated tube. It is mixed with nitrogen from a gas bottle and enters a 30 cm long test tube made of Perfluoroalkoxy alkanes (PFA) with an inner diameter (ID) of 8 mm. Behind the test tube, the gas is separated from the liquid (T03). The gas flow rate is set in three separate rotameters with different measuring ranges between 50 and 26,000 mL/min. The resistance is measured between two platinum wires submerged in the electrolyte flow at either end of the test tube with a total distance of 36 cm in between (EI01). All experiments were conducted at atmospheric vent pressure and 75 ± 5 °C electrolyte temperature. The gas is not pre-heated. Two methods are used to determine the resistance between the wires: a potentiostatic measurement at 10 V to determine the total system resistance 𝑅𝐷𝐶 including the electrochemical reaction, and a high frequency measurement at 1 - 10 kHz to determine the real part of the system's impedance 𝑍𝑟𝑒𝑎𝑙 . For the potentiostatic measurement, an assumption is made that the influence on the resistance of the reaction is small in comparison to the long electrolyte channel. The potentiostat model ZENNIUM PRO from Zahner was used for the experiments. To reduce the influence of the large fluctuations during the measurements, the shunt resistors of the potentiostat are limited to a minimum measurement range of 1.9 mA to avoid frequent shunt switching due to bubble-induced measurement instability. In addition, to minimize the variance induced by the bubbles in the high-frequency measurement, the data was reduced to only measurement points with an absolute phase shift angle |ϕ|<2° before averaging over the entire measurement. Each measurement was run at 17 points between 0 and 16,000 mL/min N2 flow corresponding to a gas fraction of 0 to 0.97 in the system. For all measurements, the set inlet gas fractions 𝜀𝑖𝑛 are listed. In addition to mixing the gases in a T-piece with 6 mm ID, which creates large bubbles similar in size to the tube diameter even at low gas flows, experiments were also carried out with a Y-mixer with 2 mm ID to study the influence of bubble size on the resistance. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 4/8 3. Results Figure 3: Standard reference case with the 8 mm ID vertical tube. Left: Comparison of the calculated resistance from the potentiostatic measurement, the impedance measurement and the filtered impedance measurement including absolute error depending on the inlet gas fraction. Right: Zoom on the comparison of the potentiostatic measurement and the filtered impedance measurement with linear y-axis. Figure 3 depicts the results of the reference case with the vertical 8 mm ID tube. On the left side, the potentiostatic measurement resistance calculated via Ohm's law 𝑅𝐷𝐶 is compared to the real high-frequency impedance 𝑍𝑟𝑒𝑎𝑙 with maximum errors. The potentiostatic resistance is calculated from the average current since the measurement's goal is to assess stray currents. In both cases, the maximum error is enormous, up to 7 orders of magnitude for the potentiostatic measurement. This variance corresponds to currents between 120 mA to 10 pA during the measurement. Effectively, the current alternates between 0 A and values in the 0.1-10 mA range during high gas fraction measurements. For the impedance measurement, the data can be similarly adjusted towards the effective current by removing all data points with high phase shift, which also make up the largest variations during the measurements. The right side of Figure 3 shows the resulting calculated resistance and filtered real impedance. They are in good agreement with each other, although the variance of the measurements increases at high gas fractions. At the highest measured gas fraction of 𝜀 = 0.97, the measured impedance is 2444 Ω compared to 63 Ω in the pure liquid. This corresponds to a reduction in current of 97 %. At 𝜀 = 0.5 the current reduction is 86 %, and a gas fraction of 𝜀 = 0.13 already achieves 51 % reduction. In addition, the influence of bubble size was investigated by comparing the standard case large T-mixer with a Y-mixer that produces significantly smaller bubbles. Pictures of the test tube at different gas fractions with a comparison of the two mixers are shown in Figure 4. The gas bubbles in the Y-mixer are much smaller than the tube diameter at small gas fractions, while the T-mixer always produces bubbles with the same diameter as the tube. In the Y-mixer, plug flow starts forming above 𝜀 =0.5 and at higher gas fractions, the difference in flow pattern becomes indistinguishable. Above 𝜀 = 0.75, the flow pattern transitions to churn flow, with the liquid becoming increasingly less continuous. An annular flow pattern where the liquid completely coats the tube surface was never observed, possibly due to the low friction of the PFA tube. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 5/8 Figure 4: Comparison of the flow patterns of the large T-Mixer (left side of each pair) and the small Y-Mixer (right side of each pair) with respective gas flow rates and inlet gas fraction. 5 cm 𝑉 𝑔𝑎𝑠 =0.05 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.09 𝑉 𝑔𝑎𝑠 = 0.1 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.17 𝑉 𝑔𝑎𝑠 = 0.3 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.38 𝑉 𝑔𝑎𝑠 = 0.7 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.58 𝑉 𝑔𝑎𝑠 = 1.5 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.75 𝑉 𝑔𝑎𝑠 = 3.0 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.86 𝑉 𝑔𝑎𝑠 = 7.0 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.93 𝑉 𝑔𝑎𝑠 =13 𝐿/𝑚𝑖𝑛 𝜀𝑖𝑛 = 0.96 5 cm EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 6/8 Figure 5: Comparison of the resistance/ impedance of the reference T-mixer to the Y-mixer and the resistance calculated by the Bruggeman correlation. Figure 5 compares the measured resistance and impedance of the two mixers. Small bubbles lead to significantly lower resistances at low gas fractions than large bubbles. The most significant measured difference has the Y-mixer at 46 % of the impedance of the Tmixer at a gas fraction of 𝜀 =0.36. When the flow pattern transitions to a churn flow above 𝜀 ≈ 0.5, the difference in impedance becomes negligible again, suggesting that the bubble size is the significant influence in this case. For comparison, the Bruggeman equation is exemplarily used as one of the common twophase conductivity correlations. Equation 1 shows the simplified equation for a gas conductivity of zero. In this case, the effective conductivity 𝜅eff only depends on the gas fraction 𝜀 and the liquid or matrix medium conductivity 𝜅m. 𝜅eff =(1−3 2𝜀)𝜅m(1) For the Bruggeman correlation in Figure 5, the matrix medium conductivity is normalized to the calculated conductivity in the pure liquid reference measurement. As expected, the equation best fits the Y-mixer conductivity with the bubbly flow, as this flow pattern violates the small particle assumption (i) of the correlation the least. For the large bubbles, the conductivity is overestimated, and the equation is not useable for high gas fractions. Although this correlation and other two-phase conductivity models have been used for gasliquid systems in the past, they are not suitable for the entire range of flow patterns common in alkaline water electrolysis systems. The experimental results presented in this work prove the assumption that gas in electrolyte channels in alkaline water electrolysis systems reduces the channel conductivity and, consequently, its stray currents. However, the amount of conductivity reduction cannot be estimated mathematically with available models and does not follow any known correlations across flow patterns. It depends not only on the gas fraction and velocity in the tube but also on the flow pattern at the inlet and the tube's size, orientation and surface. Consequently, it depends on the individual industrial electrolyzer's specific design and operating point. For example, if a higher liquid flow rate in a design also decreases the agglomeration time and thus bubble size, the impact of this change may be larger than expected. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 7/8 Additionally, while the experiments in this work were only conducted at ambient pressure, pressurized stacks will produce smaller bubbles and lower gas fractions at the same hydrogen production rate as atmospheric stacks. Accordingly, the passive stray current reduction effect at the stack outlet due to gas will be reduced in these systems. Overall, a recommendation can be made to use thin channels and large gas-to-liquid ratios in industrial systems to take advantage of this effect, allowing for the reduction of outlet channel length and pressure drop. During the presentation given at this conference, an additional comparison between a smaller tube diameter and a horizontally oriented tube will be shown. References [1] A. T. Kuhn, J. S. Booth, J Appl Electrochem. 1980, 10 (2), 233–237. DOI: https://doi.org/10.1007/BF00726091. [2] L. Chen, Y. S. Tian, T. G. Karayiannis, International Journal of Heat and Mass Transfer. 2006, 49 (21–22), 4220–4230. DOI: https://doi.org/10.1016/j.ijheatmasstransfer.2006.03.025. [3] M. Wappler, D. Unguder, X. Lu, H. Ohlmeyer, H. Teschke, W. Lueke, International Journal of Hydrogen Energy. 2022, 47 (79), 33551–33570. DOI: https://doi.org/10.1016/j.ijhydene.2022.07.253. [4] R. Qi, M. Becker, J. Brauns, T. Turek, J. Lin, Y. Song, Journal of Power Sources. 2023, 579, 233222. DOI: https://doi.org/10.1016/j.jpowsour.2023.233222. [5] A. Pellegrino, A. Culcasi, A. Cosenza, A. Cipollina, A. Tamburini, G. Micale, Chemical Engineering Science. 2024, 299, 120438. DOI: https://doi.org/10.1016/j.ces.2024.120438. [6] T. O'Brien, in Handbook of Chlor-Alkali Technology, Springer, pp. 388–398, , ISBN: 978-0-306-48624-1 2005. [7] S. Matsuda, B. P. Sullivan, MEANS FOR PREVENTING INTERNAL CURRENTS IN A FUEL CELL, US Patent 3537904, 1970. [8] F. v. Sturm, H. Cnobloch, FUEL CELLS WITH DEVICE FOR REDUCING ELECTROLYTE SHORT-CIRCUIT CURRENTS, US Patent 3522098, 1970. [9] B. Ross, S. Haussener, K. Brinkert, J. Phys. Chem. C. 2025, 129 (9), 4383–4397. DOI: https://doi.org/10.1021/acs.jpcc.5c00220. [10] R. L. Garcia Barros, J. T. Kraakman, C. Sebregts, J. Van Der Schaaf, M. T. De Groot, International Journal of Hydrogen Energy. 2024, 49, 886–896. DOI: https://doi.org/10.1016/j.ijhydene.2023.09.280. [11] R. Lira Garcia Barros, J. Scholl, I. Hoedemakers, X. L. Liang, K. Skadell, J. Van Der Schaaf, M. T. De Groot, Journal of Power Sources. 2025, 630, 236116. DOI: https://doi.org/10.1016/j.jpowsour.2024.236116. [12] R. Wu, Z. Hu, H. Zhang, J. Wang, C. Qin, Y. Zhou, Langmuir. 2024, 40 (1), 721– 733. DOI: https://doi.org/10.1021/acs.langmuir.3c02925. [13] H. Qiu, K. Obata, K. Tsuburaya, T. Nishimoto, K. Nagato, K. Takanabe, Journal of Power Sources. 2024, 611, 234765. DOI: https://doi.org/10.1016/j.jpowsour.2024.234765. [14] J. Görtz, J. Seiler, A. Jupke, International Journal of Multiphase Flow. 2024, 177, 104849. DOI: https://doi.org/10.1016/j.ijmultiphaseflow.2024.104849. [15] X. Y. Wong, Y. Zhuo, Y. Shen, Ind. Eng. Chem. Res. 2021, 60 (33), 12429–12446. DOI: https://doi.org/10.1021/acs.iecr.1c02554. [16] A. Angulo, P. van der Linde, H. Gardeniers, M. Modestino, D. Fernández Rivas, Joule. 2020, 4 (3), 555–579. DOI: https://doi.org/10.1016/j.joule.2020.01.005. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244139 A1802 Page 8/8 [17] D. Dogan, B. Hecker, B. Schmid, H. Kungl, H. Tempel, R.-A. Eichel, Electrochimica Acta. 2024, 500, 144767. DOI: https://doi.org/10.1016/j.electacta.2024.144767. [18] H. S. Burney, R. E. White, J. Electrochem. Soc. 1988, 135 (7), 1609–1612. DOI: https://doi.org/10.1149/1.2096069. [19] Markus Zahn, Patrick Grimes, Richard Bellows, Exxon Research & Engineering Co., Shunt Current Elimination and Device 4197169, US4197169, 1980. [20] H. N. Seiger, J. Electrochem. Soc. 1986, 133 (10), 2002–2007. DOI: https://doi.org/10.1149/1.2108329. [21] R. S. Jupudi, G. Zappi, R. Bourgeois, J Appl Electrochem. 2007, 37 (8), 921–931. DOI: https://doi.org/10.1007/s10800-007-9330-4. [22] S. Szpak, C. J. Gabriel, J. J. Smith, J. R. Driscoll, J. Electrochem. Soc. 1990, 137 (3), 849–853. DOI: https://doi.org/10.1149/1.2086567. [23] G. Sakas, A. Ibáñez-Rioja, S. Pöyhönen, A. Kosonen, V. Ruuskanen, P. Kauranen, J. Ahola, Renewable Energy. 2024, 225, 120266. DOI: https://doi.org/10.1016/j.renene.2024.120266. [24] G. Sakas, A. Ibáñez-Rioja, S. Pöyhönen, L. Järvinen, A. Kosonen, V. Ruuskanen, P. Kauranen, J. Ahola, Applied Energy. 2024, 359, 122732. DOI: https://doi.org/10.1016/j.apenergy.2024.122732. [25] B. Tjaden, S. J. Cooper, D. J. Brett, D. Kramer, P. R. Shearing, Current Opinion in Chemical Engineering. 2016, 12, 44–51. DOI: https://doi.org/10.1016/j.coche.2016.02.006. [26] D. A. G. Bruggeman, Annalen der Physik. 1935, 416 (7), 636–664. DOI: https://doi.org/10.1002/andp.19354160705. Keywords: EFCF2025, H2, LowTemp. Fuel Cells & Electrolysers, Stray Current, Shunt Current, Stack Design, Two-Phase Flow Remark: This work is licensed under Creative Commons Attribution 4.0 International