According to the RePowerEU plan, the yearly demand for green hydrogen production can achieve a 20 million tons level, corresponding to 660 TWh, by 2030. This will imply the connection of large electrolyser units to the grid. Such large units will greatly influence the overall system stability. Amongst the different potential related issues, one can cite the transient stability in case of close by synchronous units or maximum active power flows handling during transients. As power system studies rely heavily on the accuracy of the models used, it is crucial to ensure their representativeness. The fundamental challenge is capturing the underlying electrolyser physical processes that occur upstream, before the electrical conversion. On the one hand, electromagnetic transient (EMT) models, due to their short time step (typically in the microsecond range), can capture a wide range of rapid, short-duration events. These types of models are more likely to capture physical processes such as the electrochemical phenomena in an electrolyser cell. On the other hand, RMS models impose inherent modelling limits due to the phasor approximation, but they are more efficient for large-scale power system studies involving steady-state behaviour and longer time scales. Therefore, the objective here is to develop a phasor-based model that closely replicates the dynamics of an EMT model.

This work presents a novel, physically based RMS model of a large-scale electrolyser, developed to be used in power system stability studies. This model combines electrical and electrochemical behaviors by simulating a tap-changing transformer, a 6-pulse rectifier and an electrolyser cell. The main contributions are: (1) The implementation of a current control system based on the firing angle of the thyristors. (2) A configurable DC-side LC filter affecting the electrolyser dynamic response. (3) The inclusion of key electrochemical phenomena of the electrolyser cell, such as the reversible voltage VOC, the activation overvoltage VACT and the ohmic losses VOHM, all dynamically represented as functions of the operating current.

In this work, we first focus on the development of the proposed physical RMS model. We provide a detailed explanation about how each sub-system interacts. Then, we carry out time-domain simulations on a large-scale system with our physical-based RMS approach inspired by an EMT model. Finally, we compare our results with previous studies made on a simplified, parametric electrolyser model to highlight the importance of developing a detailed model against a simplified one. The results will be complemented by a brief literature review to highlight the limitations of existing RMS and EMT models.

The results obtained with the time-domain simulations are evaluated in terms of current, voltage and power dynamics. They are benchmarked against an EMT electrolyser model, previously validated by a manufacturer model, across several scenarios of short-circuit faults. System stability limits in 2040 prospective scenarios were evaluated with this physical-based RMS model. The simulations are performed with EUROSTAG, the simulation tool used at RTE for transient stability studies. The model will be made publicly available in the Dynaωo suite.