To reduce environmental concerns and oil dependence, hydrogen based Renewable Energy Systems (RES) appear as promising alternative for decentralized electricity production. Due to the intermittent power production of photovoltaic arrays and wind turbines, these devices are coupled with an electrolyzer to produce hydrogen when possible. Inversely, the produced hydrogen is intended to feed fuel cells, when the energy demand overtakes the energy production. This study aims to improve RES real-time diagnosis and control.

First, a dynamic model is presented to study the transient behavior of renewable energy systems. The main components of the system (photovoltaic arrays, wind turbine, batteries, electrolyzer, fuel cell and power conditioning units) were modelled individually and then collectively integrated into a global simulation model in order to operate like the real-time system. The sub-models are designed for transient and steady state voltage, current and temperature analysis. The simulation results confirm experimental measurements on the test bench. Such a global model is useful for optimal dimensioning and effective control design of the hydrogen production and utilization in RES.

Moreover, further analyses have proven that some impurities contaminate the fuel cell systems and greatly contribute to their degradation. These contaminants therefore need to be detected, and their rates be reduced in hydrogen. It is shown how the diagnosis analysis through chemical detection and atomic adsorption can help to reduce harmful compounds in the produced hydrogen. This diagnosis is a key step in the maintenance process of fuel cells in order to improve their performance, reliability and durability.