Increasing the share of renewable sources through hydrogen : optimisation and uncertainty quantification of a Power-to-Hydrogen system under stochastic conditions

As part of the fight against global warming, a general transition from fossil fuels to renewable energies is taking place. However, this is hindered by the intermittent nature of renewable energies, which limits their use. To overcome this problem, many storage methods have emerged. Amongst these solutions is hydrogen, which is particularly well known for its potential carbon neutrality and high energy density. The surplus energy generated by renewable sources can be recovered by converting it into hydrogen through the electrolysis of water via various technologies. A literature review on these different technologies was carried out to identify the most suitable one to be coupled with a volatile power source. This shows that the Proton Exchange Membrane electrolyser (PEM El) is the most suitable for dynamic operating conditions. However, the electrolyser achieves low efficiency at high load factors and suffers significant degradation under dynamic conditions. To overcome this problem, the electrolyser was coupled to batteries in order to operate only within a certain range of load factors.The batteries charge with the excess energy and discharge when the power supplied by the renewable source is low. When they run out of energy, the electricity grid supplies power to the electrolyser so that it does not shut down. In this report, the renewable energy source is simulated by a stochastic process, the Ornstein- Uhlenbeck (O-U) Geometric Brownian Motion (GBM) model. It allows to study the impact of the wind speed uncertainty on the system performance. For this purpose, the system was subjected to deterministic optimisation using a Nondominated Sorting Genetic Algorithm II (NSGA-II) with three objectives: minimising grid usage and losses when the batteries are full (grouped into a single objective), maximising the lifetime of the electrolysers and finally maximising the lifetime of the batteries. Then, an uncertainty quantification was performed on several designs with a Polynomial Chaos Expansion (PCE). The results showed that there is a real trade-off between flexibility and degradation of the electrolysers. However, a design that minimised the amount of grid and loss, with acceptable degradation, was found to be more robust than a design that minimised degradation.