Polymer electrolyte membrane fuel cells (PEMFC) are a promising technology that can significantly contribute to the reduction of greenhouse gas emissions from the mobility sector, which is still largely based on fossil fuel powered internal combustion engines.
In a PEMFC, hydrogen and atmospheric oxygen react to create water and electrical energy, the latter being usable to power an electric motor. The use of 'green' hydrogen, produced with renewable energy sources, enables emission-free power generation. Compared to other fuel cell types, PEMFCs have advantages, such as their high energy density at low operating temperatures, a high system robustness, and a quick startup capability.
However, so far high material and production costs as well as an insufficient durability have limited their broader commercial application [1]. Therefore, ongoing research is focusing on the improvement of PEMFC performance and lifetime characteristics, as well as the reduction of material and production costs [2].
Within PEMFCs bipolar plates are an important component, ensuring a sufficient supply as well as distribution of the reactant gases, removal of waste heat, and structural integrity. Coated stainless-steel foils (e.g., AISI 316L) are favoured as a base material for bipolar plates due to their numerous advantages over other materials (e.g., graphite), such as a high thermal conductivity as well as superior mechanical properties [3].
For the production of bipolar plates on an industrial scale, the metal foil is normally supplied as a coil. During the first production step, the part geometry and flow channels are introduced into the continuous metal foil during a forming process such as hydroforming or stamping. In the next step, the metal foil with the introduced geometries is separated and cut into single sheets by laser cutting or stamping, which are called bipolar half plates [4]. In order to manufacture a bipolar plate, two half plates have to be welded together in an overlap configuration.
Laser welding steps up to the plate
Laser beam welding is a state-of-the-art joining process for the production of metallic bipolar plates due to its flexibility, wear-free operation and high welding speeds. A problem during the welding process is the occurrence of process instabilities at higher welding speeds, such as the humping effect, which cause weld seam imperfections that lead to parts needing to be scrapped. An image of a weld seam with the occurrence of the humping effect was recorded with a laser scanning microscope (VK-X 1000, Keyence Corporation, Japan) and is shown in Figure 1.
Figure 1: Exemplary image of a weld seam with the occurrence of the humping effect recorded with a laser scanning microscope
In order to ensure the gas tightness of the weld seams, a leakage test is necessary. With a cycle time of 20 - 60s and the use of helium, the test is time-consuming and expensive. In total, the material and production costs of the bipolar plates are accounting for 20% of the fuel cell stack net costs per kilowatt and, thus, represent a significant share [5].
As the worldwide demand for fuel cells and therefore bipolar plates is increasing, higher welding speed, minimising process instabilities during welding, and reducing the testing time are all essential steps to lower the overall production costs and scrap rate. Hence, a reliable and advanced welding process is necessary.
Additionally, an inline process monitoring system to determine the weld seam quality is a promising approach to reduce the testing time significantly. With a mature system, it would be possible to adapt the process parameters inline and determine the gas tightness without a downstream leakage test. Furthermore, the separation of faulty bipolar plates at an earlier manufacturing stage will be possible. Both actions will help to minimise the overall production costs if realised on an industrial scale.
Process monitoring experimentation
At the Institute for Machine Tools and Industrial Management of the Technical University of Munich, the suitability of a photodiode-based sensor system (Laser Welding Monitor [LWM], Precitec, Germany) for inline process monitoring was tested. The sensor consists of three photodiodes, which measure process emissions at different wavelength ranges. In Figure 2, the signals of the three photodiodes are shown, which were recorded during the laser beam welding of two 80µm-thick AISI 316L foils. The experiments were conducted with a continuous wave multi-mode disk laser (TruDisk 1020, Trumpf, Germany) emitting at the green wavelength of 515nm.
The welding experiments were conducted using a green TruDisk 1020 laser from Trumpf
The welding speed in this example was set to 1,700mm/s to provoke the occurrence of process instabilities, such as the mentioned humping effect, and other welding defects. After processing the data, no anomaly related to the humping effect was identified within the inline measured signals. As the humping appears behind the process zone in the feed direction, the emitted radiation from this area could not be detected with the used sensor setup. Only a lack of fusion was detected based on a statistical signal analysis of the plasma signal.
Further measurements were made by recording the LWM signals during a second pass over the weld seam at the same speed but with significantly reduced laser power. In comparison to the inline measurement, the data recorded during the second pass showed clear peaks in the signal waveforms for the plasma, the temperature, and the reflection signal in the areas where humping occurred (Figure 2).
Figure 2: Measured laser welding monitor signals of a sample (P = 1,000W, v = 1,700mm/s) correlated to the corresponding weld seam
The results of the first experiments showed that a photodiode-based system can indeed detect single process defects, as a result of humping, for example, not inline but within a second pass over the weld seam. These results can contribute to a future intelligent laser beam welding architecture for the joining of metallic bipolar plates. Additionally, a sensor data fusion approach as a basis for an inline weld seam quality prediction is proposed in [6] to increase the process reliability.
The end goal of this work is to provide key information for a 100% inline quality assurance to substitute complex quality tests. The core of this system, which will be the topic of future research, is based on integrated adaptive data processing methods using machine learning algorithms for the interpretation of the sensor signals.
Christian Geiger and Tony Weiss are research associates within the Department of Laser Technologies at the Institute for Machine Tools and Industrial Management within the Technical University of Munich’s Department of Mechanical Engineering.
This article was co-authored by Michael Kick and Michael Zaeh, of the Institute for Machine Tools and Industrial Management within TUM’s Department of Mechanical Engineering.
References
[1] Li, H.; Tang, Y.; Wang, Z.; Shi, Z.; Wu, S.; Song, D.; Zhang, J.; Fatih, K.; Zhang, J.; Wang, H.; Liu, Z.; Abouatallah, R.; Mazza, A.: A review of water flooding issues in the proton exchange membrane fuel cell. Journal of Power Sources 178 (2008) 1, pp. 103–117.
[2]Geiger, C.; Kriegler, J.; Weiss, T.; Berger, A.; Zaeh, M. F.: Micro-perforation of the diffusion media for polymer electrolyte membrane fuel cells using short and ultrashort laser pulses. In: Procedia CIRP 111 (2022), pp. 796–799.
[3] Song, Y.; Zhang, C.; Ling, C.-Y.; Han, M.; Yong, R.-Y.; Sun, D.; Chen, J.: Review on current research of materials, fabrication and application for bipolar plate in proton exchange membrane fuel cell. International Journal of Hydrogen Energy 45 (2020) 54, pp. 29832–29847.
[4]Kampker, A.; Ayvaz, P.; Schön, C.; Reims, P.; Krieger G.: PRODUCTION OF FUEL CELL COMPONENTS. PEM of RWTH Aachen University and VDMA,. 1st Edition. Aachen 2020. ISBN: 978-3-947920-16-7.
[5]James, B. D.; Huya-Kouadio, J. M.; Houchins, C.: Bipolar plate cost and issues at high production rate. Southfield, Michigan; 2017.
[6]Weiss, T.; Kick, M. K.; Grabmann, S.; Geiger, C.; Mayr, L.; Wudy, K.; Zaeh, M. F.: A holistic approach for an intelligent laser beam welding architecture using machine learning for the welding of metallic bipolar plates for polymer electrolyte membrane fuel cells. Procedia CIRP 111 (2022), pp. 810–815.