Analysis of proton exchange membranes for an innovative PEM fuel cell using computational fluid dynamics (CFD)
2020
Hochschulschrift
Zugriff:
To meet the global energy demand and to maintain a green environment, there is a need for research and development into all aspect of renewable energy technology. These have hugely motivated research into the Proton Exchange Membrane (PEM) fuel cell technology. The PEM fuel cell is a clean energy technology where hydrogen electrochemically combines with oxygen to produce electricity while having heat and water as a waste product. Due to the promising nature of the technology, over 90% of fuel cell research is on PEM fuel cell [1]. In the UWS fuel cell research laboratory, there is an ongoing work to design and develop an innovative PEM fuel cell. Previous work on materials and flow plates was successful. This work focuses on the determination and development of membrane for the innovative PEM fuel cell. The fuel cell membrane is subject to various operating condition with the best performance adopted for the innovative PEM fuel cell. This work is in 3 different segments; ▪ Modelling and simulation using ANSYS ▪ Experimental fuel cell testing and ▪ development of predictive models using supervised classical machine learning models. The role of computational fluid dynamics (CFD) in fuel cell research, especially in the modelling and simulation is enormous. This software used in this work is ANSYS Fluent. The analysis included varying the operating condition of the fuel cell, such as operating temperature, operating pressure, and water inlet at the cathode side. There was a comparison between the performance of membranes having different thicknesses. In furtherance to this, there was a PEM fuel cell testing experiment. In the experiment, voltages were set to 0.5V, 1.0V, 1.5V, 2.0V and 2.5V respectively, and corresponding results, recorded. There was a comparison of the experimental and simulation results and validation using statistical machine learning methods. Best results were at 2.0V and 2.5V. The worst result is at 0.5V. There were limited variations in the performance at voltages 2.0V and 2.5V, therefore recommended voltage for the innovative fuel cell is 2.0V. There was a comparison of performance at temperature 333K, 343K and 353K. The best performance in all the various experiments was recorded at 353K and was, therefore, recommended. The polarisation curve result showed that the thinner membranes of 50.8μm performed better than the 127μm membrane. Besides, the thickness shows more influence on the performance of the innovative PEM fuel cell than temperature. They both perform well when subjected to the right operating conditions — the comparison between the simulation and experimental results. The results followed the same pattern showing a high level of agreement. Statistical machine learning tools - Python on Jupyter notebook was in use for predictive modelling. It produced a training accuracy of 98% and testing accuracy of 94% using regression. The Mean square error (MSE) for the training and testing set is 0.0015 and 0.0035, respectively. Nine supervised classification models developed, trained and tested produced the MPL classifier as the best result after cross-validation.
Titel: |
Analysis of proton exchange membranes for an innovative PEM fuel cell using computational fluid dynamics (CFD)
|
---|---|
Autor/in / Beteiligte Person: | Ogungbemi, Emmanuel Olubunmi |
Link: | |
Veröffentlichung: | 2020 |
Medientyp: | Hochschulschrift |
Sonstiges: |
|