This work explores the implementation of coordinated control for a portfolio of electrolyzers.

The Power-to-X technologies are expected to become an important part of Offshore Energy Hubs (OEH), offering demand-side flexibility, optimizing wind energy production and providing power balancing services. The latter implies the implementation of a control strategy for a large-scale electrolysis plant, which can consist of a variety of electrolyzers with different flexibility capabilities. The identical control strategy implemented for electrolyzers with different parameters within one electrolysis plant can lead to non-optimal operation of electrolyzer and can cause negative consequences, such as power oscillations between the two types of electrolyzers.

OEH, with 2 HVDC lines with 2 GW capacity each, 4 Wind Power Plants (WPP) with 1 GW capacity each and a 1 GW electrolysis plant built and simulated in PowerFactory. Electrolysis plant includes a portfolio of 4 electrolyzers from different manufacturers, meaning they have different physical and safety limitations (e.g. ramp-rate limit, minimum permitted power). Each electrolyzer is connected to the hub via a power supply unit with the power response scheme, such as a droop controller and Automatic Generator Controller (ACG).

The methodology used for electrolyzer modelling combines empirically estimated voltage-current characteristics for steady-state operation, a generic equivalent circuit representation to capture dynamic response, and control limitations related to plant-level ramp-rate restrictions.

The system is tested under two main fault scenarios. First, a fault disconnects the offshore wind power plants, causing the system to experience under-frequency. Second, an HVDC converter pole fault is considered, leading to an over-frequency situation.

Under these fault scenarios, several approaches for coordinated control were considered. For the base-case scenario, all electrolyzer were equipped with the same ACG. Then, the droop controller was enabled on more flexible electrolyzers, making the response faster. Finally, for under-frequency events, the procedure of switching to stand-by mode was implemented based on the electrolyzer characteristics, specifically, on the time needed to restore hydrogen production after the fault is cleared in order to minimize time of restoration.

The preliminary conclusions are the following: for the base case with identical control settings on electrolyzers with different parameters, the frequency power oscillations were observed due to “hunting” between slow and fast electrolyzers. By coordinating electrolyzers’ response (e.g droop controller or AGC), it is possible to minimize power curtailment and avoid frequency overshoot due to assigning frequency restoration only to one of the electrolyzer groups.

This work results and conclusions are based on the master's thesis work.