A Sensitivity Study of Key Parameters on the Efficiencies and Energy Losses of Organic Redox Flow Batteries
Yunxiang Chen,Chao Zeng,Yucheng Fu,Jie Bao,Zhijie Xu,Peiyuan Gao,Soowhan Kim,Wei Wang
DOI: https://doi.org/10.1149/ma2023-013736mtgabs
2023-08-28
ECS Meeting Abstracts
Abstract:Organic redox flow batteries (ORFB) are promising cost-competitive technologies for large-scale energy storage. At the unit cell scale, the battery performance is controlled by the combined effects of key parameters of microstructure design, physicochemical properties of organic molecules, mass transport, operating conditions, and membrane properties. However, a comprehensive study of how these parameters could affect the overall performance of ORFB has not been done. To address this gap, this work presents a sensitivity study of 12 key parameters on the battery efficiencies and energy losses of ORFB by using a validated numerical model for the Dihydroxyphenazine (DHP) family. The 12 key parameters include the specific area of microstructure; mass transfer rate; diffusivity, viscosity, reaction transfer coefficient, the reaction rate constant, standard potential, and electronic conductivity of DHP anolyte; operation conditions such as initial concentration of active species, pump rate, and applied current; and membrane electronic conductivity. In total, we generated 15,360 combinations of the 12 key parameters. These data are split into 15 groups with each having a fixed current density and flow rate and 1024 random parameter combinations of the remaining 10 parameters. All these parameters are then sent to the validated ORFB model developed in Comsol to predict the coupled flow, reaction, and transport processes at 38 pre-defined SOC between 0.025 and 0.85. We then output the total potential, equilibrium potential, activation potential, and concentration potential for charge (19 SOC) and discharge (19 SOC) from the model to further calculate the Coulombic Efficiency, Energy Efficiency, Voltage Efficiency, and System Efficiency as well as the energy losses contributed by equilibrium, activation, concentration, and ohmic losses. With these calculated efficiencies and the relative importance of energy losses, we used a deep neural network (DNN) to fit the relationship between the input 12 parameters (15,360 combinations), and the model calculated 4 efficiencies/relative ratios of energy losses. A sensitivity score is then calculated based on the DNN model and key factors controlling the ORFB performance are determined.