(399d) Development of Condition Monitoring and Prognostic Capabilities for a Vanadium Redox Flow Battery

The vanadium redox flow battery (VRFB) is a rechargeable flow battery that is one of the most promising large-scale energy storage systems making them suitable for grid energy storage. However, the self-discharge reactions along with the undesired side reactions and water transfer through the membrane causes imbalance in electrolyte and state-of-charge (SOC), which can significantly reduce the capacity of VRFBs. Self-discharge reactions, caused by the diffusion, convection, and migration of the vanadium ions from one half-cell to the other, lead to an imbalance between the state-of-charge of the two half-cell electrolytes and consequently cause a capacity fade. As a result of evolution of hydrogen or air oxidation of V²⁺, the side reactions also affect the capacity of the VRFB. Likewise, inadequate flow rate reduces the cycling time, thus affecting the capacity. To overcome the effect of side reactions, it has been suggested that the operating range of the cell be maintained between 10% and 90% SOC^[1]. However, even when the overall SOC is constrained in this desired region, there can be considerable local inhomogeneity causing capacity loss. For maximizing the reliability of VRFBs under load-following applications, capacity fade needs to be estimated along with a prognostic capability that can estimate the remaining useful life (RUL) of the battery.

A first-principle non-isothermal model^[2] is developed based on the physical processes and electro-chemical reactions taking place inside a cell to capture the dynamics of the model. This model is based on momentum and mass transfer, ion conservation, and energy transport phenomena. This is combined with a kinetic model for electro-chemical reactions involving vanadium species. Bubble formation due to the gas evolving reactions at the positive and negative electrodes is included to account for the effect of momentum transfer between gas and liquid phases and also the reduction in liquid volume. Heat generation due to activation losses, electrochemical reaction, and Ohmic resistance is modeled. Water transport through the membrane is also modeled. Due to the different diffusion coefficients of the vanadium species, their cross-over causes capacity decay. In addition, the gas evolving reactions are modeled. These reactions lead to an imbalance in the SOC leading to capacity loss.

A model-based approach is developed for condition monitoring and prognosis. The detailed first-principles model is computationally intractable for prognosis. Thus, an adaptive neural network (NN) model-based approach is developed to estimate SOC and capacity fade. Given the current state and quantified uncertainty in the NN model and model parameters, the probability density function (PDF) for the RUL is calculated. The PDF for the RUL is conditional and is updated at discrete time intervals taking into consideration the accumulated damage and expected future operating profile.

Reference

1. Tang, Ao, Bao, Jie, and Maria Skyllas-Kazacos, “Dynamic modelling of the effects of ion diffusion and side reactions on the capacity loss for vanadium redox flow battery”, Journal of Power Sources 196 (2011) 10737– 10747.
2. Hassan Al-Fetlawi. “Modelling and Simulation of All-Vanadium Redox Flow Batteries”. Thesis for the degree of Doctor of Philosophy. University of Southampton. November 2010.