As a starting point, fuel cell stack components aged in real automotive conditions have been identified from previous projects for post-mortem analyses. Reference ageing tests in real conditions were performed on test stations to get exploitable reference data with state of art MEA components and stacks. A new realistic protocol based on actual fleet data, has been defined as ID-FAST driving cycles, with specific conditions and various stop-events mimicking real life operation for an automotive application.
Real ageing testing plan in single cells and stacks provided information on stressors. Valuable results with ranking of MEAs for several designs could be obtained thanks to different core materials, with at least two types of CCMs and GDLs selected according to the stack designs. Results in stacks showed how the robustness of MEAs and homogeneity between cells are crucial criteria for correct durability assessment. Analyses by ex-situ methods showed following degradation issues: cathode platinum particle dissolution, carbon corrosion when hydrogen/air front occurred, hydrophobicity loss of GDLs, gas leakages due to membrane failure in hot-dry operation.
Main mechanisms have been modelled for the core materials. Modelling work for the development of ASTs started with the analysis of stressors like for the carbon corrosion mechanism related to the start-up / shutdown steps. Catalyst AST simulations confirmed the pertinence of the approach by showing better representativeness. Progress was done on the coupling of physical models describing platinum catalyst and membrane degradation. Simulation results were obtained by a new approach coupling membrane chemical degradation and metallic cations coming from the possible corrosion of bipolar plates’ material.
An extensive modelling study on the GDLs allowed improving their description with the effect of composition, and structure of support and microporous layer. A parametric study showed the excess of temperature as the best basis for a new AST on GDLs.
However, all first results confirmed that the degradation of cathode catalyst layers was leading the performance losses at least for ageing cases and timeframe considered.
Definition of operando single components AST and AST with multiple stressors based on drive cycles could be really started after mid-term. First new results were on parameters affecting main mechanisms such as local cathode conditions on platinum dissolution or potential cycling on performance decay. The development of combined ASTs could be completed by following the selected approach consisting in modifying the ID-FAST driving cycles. The idea was to preserve but aggravate the stressors and mechanisms involved, meanwhile keeping the original design of cycles with low power and high power periods, aiming at reaching the same final performance in a shorter duration.
Stronger accelerating effect was observed with the Low Power / high Power AST cycles deeply investigated at single cell level. These ASTs were first tested separately then combined in alternating blocks of 200 cycles of each, leading to acceleration factors of 7 to 10 in terms of voltage decay rates, and to a transfer function, simpler than expected, being a ratio of 1 to 1 for the number of cycles. For the exacerbated load cycles, with lower minimum load and higher current peaks compared to the reference, tests accelerated active area losses and voltage decay but acceleration factors were smaller. Reducing slightly dwell times and enhancing some stressors, could help improving acceleration with robust MEAs. All combined AST protocols could be transferred to stack scale. If overall behaviour and ranking of MEAs are similar with different designs, acceleration factors were different from those assessed in single cells, probably due to large size effect not fully resolved.
Thanks to more experimental ageing data, successful simulation of cycles for both real ageing and ASTs could be conducted. With modelling of the differential cell performance, simulation of low power AST could be done confirming the additional impact of air stops. For the automotive stack design, simulation of the whole cell highlighted the temperature stressor effect during the high power part of cycles. Simulation of ID-FAST cycles, up to 1000 for the reference, involving the platinum Ostwald ripening mechanism, showed the accelerating effect of increasing temperature and strong impact of reducing minimum load and increasing voltage amplitude.
As a general conclusion, despite issues encountered for the availability of appropriate components and test objects and additional time lost due to covid-19 situation, ID-FAST could successfully achieve expected outcomes. Combined AST protocols were indeed developed and validated with regard to their relevance and their capability to actually reduce testing time compared to real ageing.
Key exploitable results identified by the consortium: ID FAST driving cycle (based on BMW customer data) verified and adapted to commonly used test stations; validated ASTs; methodology and protocols for AST dedicated to PEMFC, Polimi’s Zero-P and multi-zero-P (four single cell tests in parallel) hardware.
Results have been disseminated in scientific conferences, during the successful ID-FAST final workshop and with 11 open-access publications.
Recommendations are available as support to standardization activities on AST for Fuel Cells at the IEC TC105.
