Abstract

Redox flow batteries (RFBs) are regarded as one of the most promising grid-scale energy storage technologies, which can improve the applicability of wind and solar energy by leveling their fluctuating power output. In RFBs, the gas side reactions including hydrogen evolution reaction and oxygen evolution reaction reduce the efficiency and capacity of the batteries and the gas produced may even cause safety problems. Even if the voltage is controlled in practical applications to restrict the gas side reactions, it is still inevitable due to flow dead zones or insufficient reactants locally inside the electrode. Gas side reactions are usually caused by too high local overpotential, i.e., local over polarization. Therefore, in this work, we propose a multi-scale model that allows the rapid prediction of local polarization to limit and manage the gas side reactions. The multi-scale model contains a deep neural network, a pore network model, and a three-dimensional continuum model with the advantages of both accuracy and extensibility. By learning from the sample data provided by the cell-scale model and the pore-scale model, the deep neural network can quickly predict the local polarization under different operating conditions and give the distribution map of the entire electrode. Compared with conventional models, multi-scale models can take into account the geometry of electrode fibers and are suitable for electrode design and modification. Due to the calculation superiority of the pore network model, the multi-scale model can realize three-dimensional local polarization prediction, which enables a more accurate analysis of the internal uniformity of the electrode. Through the developed model, we explore the effects of the interdigitated flow field, flow rate, initial concentration and voltage on local polarization. It is shown that the multi-scale model can accurately predict local polarization. However, the promotion of flow rate cannot completely eliminate the occurrence of local over polarization.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call