Abstract:
The performance improvement of a vanadium redox flow battery (VRFB) was focused on this study. The two objectives of this study were (1) to investigate the effect of the operating temperatures of the Atmospheric Pressure Plasma jets (APPJs) process on the energy efficiency of the VRFB and (2) to determine the optimal electrolyte flow rate of the VRFB by solving a dynamic optimization based on a neural network model of the VRFB. The APPJs graphite felt electrode treatment temperature providing the highest energy efficiency of the VRFB was 550°C, explained by the Energy Dispersive X-ray Spectrometry (EDX) and X-ray photoelectron spectroscopy (XPS) results. The EDX results indicated that the electrode treated with APPJs at 550°C had a high percentage of the oxygen atom, and XPS results illustrated the highest C=O functional group on the surface of APPJs at 550°C electrode comparing to the APPJs at other temperatures and sulfuric treatment. Moreover, the wettability of the electrode with APPJs treatment at all temperatures was higher improved than that with a sulfuric acid treatment and an untreated one. However, the electrode treatment did not visibly change the surface of the electrode, as shown in Scanning Electron Microscopy (SEM) results. In the second part of this study, an optimization of the electrolyte flow rate based on a neural network (NN) model of the VRFB was investigated. The NN was separately trained between the charging and discharging process using a nonlinear autoregressive with external input (NARX) model. The training and testing results indicated a high accuracy of the NN model. Under the optimal electrolyte flow rate obtained by solving the NN based-optimization, the VRFB provided a high system efficiency (SE) due to reducing the concentration overpotential. It demonstrated that the NN model could replace the theoretical model used in the dynamic optimization, even though the performance of the VRFB under the optimal electrolyte flow rate obtained from the NN-based optimization is slightly lower than that from the theoretical model-based optimization.
