Organic flow-battery lifetime prediction

A flow battery stores energy in liquids pumped through an electrochemical cell. Can we forecast permanent loss of its active material when the temperature or charging schedule changes? Organic molecules in those liquids change between charged and discharged forms. Some molecules also undergo unwanted reactions. A battery kept charged overnight may therefore lose usable material even if no charge–discharge cycle is completed.[1] [2]

The battery’s liquid can lose capacity because active molecules are destroyed, or because its two sides become unbalanced. Can a reaction-based model distinguish those outcomes and forecast permanent loss under a different charging schedule or temperature? In a study of the quinone BQDS, light-absorption measurements made during operation were combined with multivariate curve resolution and Bayesian inference to separate reactants and products and estimate uncertain rates. Water-driven Michael addition to oxidized quinone species explained the measured chemical loss; overlapping product spectra left some parameters less certain. That mechanism is demonstrated progress within the studied conditions, while transferable lifetime prediction remains the target.[1] [2]

Researchers can identify particular decay reactions and estimate their rates. The unresolved task is predicting permanent loss when operating conditions change, while separating it from capacity that can be recovered by rebalancing the liquids. A curve fitted to one cycling schedule does not establish that forecast.[1] [2]

Its validated experiments used dilute solutions, while practical organic electrolytes may be much more concentrated. Transport, crossover and several reactions can interact under changing loads and temperatures. The remaining forecast must test how inferred rates and species assignments transfer to those conditions, distinguish recoverable imbalance from molecular destruction, and attach uncertainty to predictions made beyond the calibration history.[1] [2]

A maintenance decision depends on the cause of the loss. Rebalancing can correct some inventory differences between the two sides, but cannot recreate molecules destroyed by chemistry. Forecasting those two outcomes separately would help determine which operating schedule preserves useful energy storage and when intervention can actually recover it.[1] [2]

This problem combines observability with model transfer. If different chemical states produce nearly indistinguishable spectra or capacity curves, additional measurements may be needed to identify the reaction network. Once identified, that network still needs an operating-history test: agreement with the experiment used to infer its rates is weaker evidence than correctly forecasting a different standby and cycling schedule.[1] [2]