Impedance spectroscopy techniques for condition monitoring of polymer electrolyte membrane fuel cells

Energy continues to remain the spine of all human development. As we continue to make advances in various levels, the need for energy in quantity, and even more recently, quality, continues to increase. The fuel cell presents itself as a promising prospect to solve one of mankind’s current challenge - clean energy. The fuel cell is essentially an electrochemical conversion system which takes in fuel supply to produce electricity. Some key features make the fuel cell attractive as a power source. Firstly, its efficiency in practical applications is approximately 50% compared to the typical efficiency of 40% for a typical internal combustion engine [1]. Secondly, unlike the systems such as the internal combustion engine that typically releases carbon-monoxide which is a major greenhouse gas, the typical fuel cell system, produces just water and heat, alongside the useful electrical energy. These characteristics make it attractive as a clean energy supply capable of replacing the fossil-based supplies that are currently the mainstay. Unfortunately, the fuel cell is far cry from an ideal system. Despite significant advantages of the fuel cell as a power supply, various challenges still exist which have hindered its widespread acceptance and deployment. The fuel cell at its core is a highly multi-physics system and its operational intricacies makes it highly prone to a series of fault conditions. This begs the question of durability - an important requirement of a viable power source. Another challenge is the fact that humanity currently struggles with an efficient method of producing hydrogen which is the fuel of choice for the fuel cell. Given the promises of the fuel cell however, research efforts continue to increase to further improve its viability as an energy source competitive enough to meet mankind’s need of clean energy. This work presents results bordering on efficient diagnostic approaches for the fuel cell, aimed at improving the durability of the fuel cell. Particularly, two techniques targeted at improving the popular Electrochemical Impedance Spectroscopy (EIS) are presented. Conventional EIS takes significant amount of time, rendering it unsuitable for real-time diagnostics. Multi-frequency perturbation signals have been proposed to address this challenge. These however introduces concerns surrounding the accuracy of the resulting impedance measurement. Part of this work addresses some of the challenges with the fuel cell multi-sine impedance spectroscopy, such as measurement accuracy, by defining an optimized signal synthesis formulation. The proposed approach is validated in simulation and compared to the popular exponential frequency distribution approach using the appropriately defined error metric. Secondly, the chirp – as a frequency rich signal, is investigated as an alternative perturbation signal. Consequently, the use of the wavelet transform as an analysis tool of choice is presented. The characteristic nature of the chirp signal makes a broadband frequency sweep over time possible, hence enabling a faster impedance estimation. The resulting decomposition is harnessed for impedance calculation. The approach is tested in simulation and results for equivalent circuits are presented. It is shown that the resulting impedance spectrum well approximates the theoretical values. To further validate both techniques in practice, a low-cost active load is designed and built. The active load enables the injection of an arbitrary signal using the load modulation technique. The device is tested and benchmarked against commercial frequency response analyzer (FRA) using the conventional single sine EIS technique. Both approaches developed – the improved multi-sine scheme and the chirp signal perturbation are demonstrated with the aid of the active load on a single cell fuel cell station. Outcomes of the experiment show significant accuracy from the two techniques in comparison with results obtained from the FRA equipment which implements the single sine technique. In addition, the two schemes enabled impedance results to be taken in a few seconds, compared to conventional single sine EIS which takes several minutes. Impedance measurements are also carried out in the presence of two prominent faulty conditions – flooding and drying, using the developed techniques. This demonstrates the capability of the proposed system to perform real-time diagnostics of the PEMFC using impedance information