A flexible analytical model for operational investigation of solar hydrogen plants

Abstract

Hydrogen will become a dominant energy carrier in the future and the efficiency and lifetime cost of its production through water electrolysis is a major research focus. Alongside efforts to offer optimum solutions through plant design and sizing, it is also necessary to develop a flexible virtualised replica of renewable hydrogen plants, that not only models compatibility with the “plug-and-play” nature of many facilities, but that also identifies key elements for optimisation of system operation. This study presents a model for a renewable hydrogen production plant based on real-time historical and present-day datasets of PV connected to a virtualised grid-connected AC microgrid comprising different technologies of batteries, electrolysers, and fuel cells. Mathematical models for each technology were developed from chemical and physical metrics of the plant. The virtualised replica is the first step toward the implementation of a digital twin of the system, and accurate validation of the system behaviour when updated with real-time data. As a case study, a solar hydrogen pilot plant consisting of a 60 kW Solar PV, a 40 kW PEM electrolyser, a 15 kW LIB battery and a 5 kW PEM fuel cell were simulated and analysed. Two effective operational factors on the plant's performance are defined: (i) electrolyser power settings to determine appropriate hydrogen production over twilight periods and/or overnight and (ii) a user-defined minimum threshold for battery state of charge to prevent charge depletion overnight if the electrolyser load is higher than its capacity. The objective of this modelling is to maximise hydrogen yield while both loss of power supply probability (LPSP) and microgrid excess power are minimised. This analysis determined: (i) a hydrogen yield of 38–39% from solar DC energy to hydrogen energy produced, (ii) an LPSP <2.6 × 10−4 and (iii) < 2% renewable energy lost to the grid as excess electricity for the case study.