Lithium-ion (Li-ion) batteries are a promising source of electrical energy storage due to their improved energy and power densities coupled with potential cost savings compared to previous battery chemistries. However, significant research is needed to achieve a level of technical maturity that enables greater market penetration in the electrified vehicle segment. Energy density is currently an opportunity for improvement, and this shortcoming is compounded by the unavoidable aging process that shortens battery life by reductions in the energy and power that a battery can deliver.
Model-based analysis may be used to identify and suggest mitigation strategies for the performance limiting factors. Most battery models are macroscopic and ignore the presence of interfaces between the solid and liquid phases. In these regions, known as the electrical double layer, the ionic concentration and potential vary rapidly. A better understanding of the electrical double layer leads to improved models for interfacial charge transfer. The potential variation within the electrical double layer may also modify the rate of side reactions that occur in close proximity to the electrode surface, when compared with macroscopic models.
Model order reduction techniques applied to the partial differential equations of porous electrode theory leads to models that can be used for parameter estimation and large-scale aging simulations, but retain important aspects of electrochemistry. Since the developed models of lithium ion transport and potential variation across the battery unit cell are of low computational order, parameter estimation techniques may be incorporated to gain insight into the set of parameters that represent aging. Application of nonlinear least squares estimation is particularly powerful because the models exhibit dependence on electrochemical parameters that have physical meaning. Two case studies are presented for the reduced order modeling techniques that incorporate chemistry-specific phenomena.
Model-based diagnostics are useful to understand the aging process, since in situ methods for measuring the aging process are often not feasible due to the small spatial dimensions and long time scales involved. Diagnostic methods are applied to characterize the process of capacity loss for the two studied cell types. Once the performance limiting mechanisms are understood, predictive models can be developed. To address the instance where loss of cyclable lithium is deemed the dominant capacity fade mode, a capacity fade model is developed based on a novel interpretation of sold-electrolyte interphase (SEI) layer growth.
This dissertation contains the development of reduced-order models suitable for aging parameter estimation, an identification of the dominant capacity fade mechanisms via a model-based analysis for two types of commercially available Li-ion cells, a micro-scale model of the electrical double layer near each electrode, and a novel model of SEI growth. In future work, the SEI growth model can be integrated with improved understanding of the electrical double layer to provide high fidelity capacity fade prediction.